{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 10.446343779677113,
  "global_step": 11000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7,
      "compression_loss": 0.0,
      "distillation_loss": 1.842069387435913,
      "epoch": 0.0,
      "learning_rate": 5e-05,
      "loss": 1.7228,
      "step": 1,
      "task_loss": 0.6496906280517578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7000356083082708,
      "compression_loss": 0.0,
      "distillation_loss": 1.8232104778289795,
      "epoch": 0.0,
      "learning_rate": 5e-05,
      "loss": 1.6991,
      "step": 2,
      "task_loss": 0.5824222564697266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7000712081625273,
      "compression_loss": 0.0,
      "distillation_loss": 1.8167695999145508,
      "epoch": 0.0,
      "learning_rate": 4.99999977293148e-05,
      "loss": 1.6968,
      "step": 3,
      "task_loss": 0.6166934967041016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7001067995637733,
      "compression_loss": 0.0,
      "distillation_loss": 1.810304880142212,
      "epoch": 0.0,
      "learning_rate": 4.999999091725961e-05,
      "loss": 1.6919,
      "step": 4,
      "task_loss": 0.6265115737915039
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7001423825130124,
      "compression_loss": 0.0,
      "distillation_loss": 1.7883914709091187,
      "epoch": 0.0,
      "learning_rate": 4.999997956383567e-05,
      "loss": 1.6653,
      "step": 5,
      "task_loss": 0.5578117370605469
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7001779570112483,
      "compression_loss": 0.0,
      "distillation_loss": 1.4738185405731201,
      "epoch": 0.01,
      "learning_rate": 4.999996366904504e-05,
      "loss": 1.3741,
      "step": 6,
      "task_loss": 0.47693514823913574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7002135230594846,
      "compression_loss": 0.0,
      "distillation_loss": 1.6521902084350586,
      "epoch": 0.01,
      "learning_rate": 4.999996366904504e-05,
      "loss": 1.5516,
      "step": 7,
      "task_loss": 0.646312952041626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.700249080658725,
      "compression_loss": 0.0,
      "distillation_loss": 1.5596864223480225,
      "epoch": 0.01,
      "learning_rate": 4.999994323289061e-05,
      "loss": 1.4611,
      "step": 8,
      "task_loss": 0.5743134021759033
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7002846298099733,
      "compression_loss": 0.0,
      "distillation_loss": 1.4804778099060059,
      "epoch": 0.01,
      "learning_rate": 4.999991825537609e-05,
      "loss": 1.3854,
      "step": 9,
      "task_loss": 0.5301415920257568
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7003201705142331,
      "compression_loss": 0.0,
      "distillation_loss": 1.4417721033096313,
      "epoch": 0.01,
      "learning_rate": 4.999988873650602e-05,
      "loss": 1.343,
      "step": 10,
      "task_loss": 0.4539656639099121
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.700355702772508,
      "compression_loss": 0.0,
      "distillation_loss": 1.2685056924819946,
      "epoch": 0.01,
      "learning_rate": 4.999985467628575e-05,
      "loss": 1.1892,
      "step": 11,
      "task_loss": 0.47563648223876953
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7003912265858019,
      "compression_loss": 0.0,
      "distillation_loss": 1.2336152791976929,
      "epoch": 0.01,
      "learning_rate": 4.999981607472149e-05,
      "loss": 1.1549,
      "step": 12,
      "task_loss": 0.44687366485595703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7004267419551183,
      "compression_loss": 0.0,
      "distillation_loss": 1.1893235445022583,
      "epoch": 0.01,
      "learning_rate": 4.999977293182023e-05,
      "loss": 1.1171,
      "step": 13,
      "task_loss": 0.4666560888290405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7004622488814609,
      "compression_loss": 0.0,
      "distillation_loss": 1.2539210319519043,
      "epoch": 0.01,
      "learning_rate": 4.999972524758982e-05,
      "loss": 1.1771,
      "step": 14,
      "task_loss": 0.48534464836120605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7004977473658335,
      "compression_loss": 0.0,
      "distillation_loss": 0.7343780994415283,
      "epoch": 0.01,
      "learning_rate": 4.999967302203893e-05,
      "loss": 0.6899,
      "step": 15,
      "task_loss": 0.28955867886543274
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7005332374092396,
      "compression_loss": 0.0,
      "distillation_loss": 1.198635220527649,
      "epoch": 0.02,
      "learning_rate": 4.9999616255177016e-05,
      "loss": 1.1238,
      "step": 16,
      "task_loss": 0.4500184953212738
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.700568719012683,
      "compression_loss": 0.0,
      "distillation_loss": 0.9082566499710083,
      "epoch": 0.02,
      "learning_rate": 4.999955494701443e-05,
      "loss": 0.851,
      "step": 17,
      "task_loss": 0.3353431224822998
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7006041921771674,
      "compression_loss": 0.0,
      "distillation_loss": 1.1613315343856812,
      "epoch": 0.02,
      "learning_rate": 4.999948909756227e-05,
      "loss": 1.0989,
      "step": 18,
      "task_loss": 0.5371114015579224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7006396569036965,
      "compression_loss": 0.0,
      "distillation_loss": 1.3335918188095093,
      "epoch": 0.02,
      "learning_rate": 4.9999418706832525e-05,
      "loss": 1.2628,
      "step": 19,
      "task_loss": 0.626000165939331
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7006751131932738,
      "compression_loss": 0.0,
      "distillation_loss": 1.0266211032867432,
      "epoch": 0.02,
      "learning_rate": 4.9999343774837976e-05,
      "loss": 0.9634,
      "step": 20,
      "task_loss": 0.39482954144477844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7007105610469031,
      "compression_loss": 0.0,
      "distillation_loss": 0.8103340864181519,
      "epoch": 0.02,
      "learning_rate": 4.999926430159223e-05,
      "loss": 0.7583,
      "step": 21,
      "task_loss": 0.2896346151828766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7007460004655882,
      "compression_loss": 0.0,
      "distillation_loss": 1.3146494626998901,
      "epoch": 0.02,
      "learning_rate": 4.9999180287109725e-05,
      "loss": 1.2389,
      "step": 22,
      "task_loss": 0.5570744276046753
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7007814314503326,
      "compression_loss": 0.0,
      "distillation_loss": 0.8772894740104675,
      "epoch": 0.02,
      "learning_rate": 4.999909173140572e-05,
      "loss": 0.8244,
      "step": 23,
      "task_loss": 0.3481764495372772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.70081685400214,
      "compression_loss": 0.0,
      "distillation_loss": 0.7861129641532898,
      "epoch": 0.02,
      "learning_rate": 4.999899863449631e-05,
      "loss": 0.736,
      "step": 24,
      "task_loss": 0.2852080166339874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7008522681220143,
      "compression_loss": 0.0,
      "distillation_loss": 1.023297667503357,
      "epoch": 0.02,
      "learning_rate": 4.99989009963984e-05,
      "loss": 0.9664,
      "step": 25,
      "task_loss": 0.45472821593284607
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7008876738109588,
      "compression_loss": 0.0,
      "distillation_loss": 0.6092305183410645,
      "epoch": 0.02,
      "learning_rate": 4.999879881712973e-05,
      "loss": 0.5785,
      "step": 26,
      "task_loss": 0.3023172616958618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7009230710699776,
      "compression_loss": 0.0,
      "distillation_loss": 0.6762804985046387,
      "epoch": 0.03,
      "learning_rate": 4.999869209670885e-05,
      "loss": 0.6369,
      "step": 27,
      "task_loss": 0.28206872940063477
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7009584599000741,
      "compression_loss": 0.0,
      "distillation_loss": 1.0091867446899414,
      "epoch": 0.03,
      "learning_rate": 4.999858083515517e-05,
      "loss": 0.9509,
      "step": 28,
      "task_loss": 0.4264417588710785
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7009938403022521,
      "compression_loss": 0.0,
      "distillation_loss": 0.5569751262664795,
      "epoch": 0.03,
      "learning_rate": 4.999846503248888e-05,
      "loss": 0.5272,
      "step": 29,
      "task_loss": 0.2587721347808838
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7010292122775152,
      "compression_loss": 0.0,
      "distillation_loss": 1.1096380949020386,
      "epoch": 0.03,
      "learning_rate": 4.9998344688731027e-05,
      "loss": 1.0493,
      "step": 30,
      "task_loss": 0.5061087608337402
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7010645758268672,
      "compression_loss": 0.0,
      "distillation_loss": 1.5996776819229126,
      "epoch": 0.03,
      "learning_rate": 4.999821980390346e-05,
      "loss": 1.5213,
      "step": 31,
      "task_loss": 0.8160263299942017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7010999309513116,
      "compression_loss": 0.0,
      "distillation_loss": 1.0707523822784424,
      "epoch": 0.03,
      "learning_rate": 4.999809037802888e-05,
      "loss": 1.0159,
      "step": 32,
      "task_loss": 0.5219181776046753
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7011352776518522,
      "compression_loss": 0.0,
      "distillation_loss": 0.9755439162254333,
      "epoch": 0.03,
      "learning_rate": 4.999795641113079e-05,
      "loss": 0.9262,
      "step": 33,
      "task_loss": 0.4821498394012451
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7011706159294928,
      "compression_loss": 0.0,
      "distillation_loss": 1.1395909786224365,
      "epoch": 0.03,
      "learning_rate": 4.9997817903233527e-05,
      "loss": 1.1089,
      "step": 34,
      "task_loss": 0.8322716355323792
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7012059457852369,
      "compression_loss": 0.0,
      "distillation_loss": 0.9883664846420288,
      "epoch": 0.03,
      "learning_rate": 4.999767485436224e-05,
      "loss": 0.9305,
      "step": 35,
      "task_loss": 0.41015535593032837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7012412672200883,
      "compression_loss": 0.0,
      "distillation_loss": 0.8167462944984436,
      "epoch": 0.03,
      "learning_rate": 4.999752726454293e-05,
      "loss": 0.7736,
      "step": 36,
      "task_loss": 0.3853650391101837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7012765802350505,
      "compression_loss": 0.0,
      "distillation_loss": 0.7159909009933472,
      "epoch": 0.04,
      "learning_rate": 4.9997375133802415e-05,
      "loss": 0.6719,
      "step": 37,
      "task_loss": 0.27524322271347046
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7013118848311274,
      "compression_loss": 0.0,
      "distillation_loss": 0.8945153951644897,
      "epoch": 0.04,
      "learning_rate": 4.999721846216831e-05,
      "loss": 0.8469,
      "step": 38,
      "task_loss": 0.4183288812637329
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7013471810093225,
      "compression_loss": 0.0,
      "distillation_loss": 0.7273141145706177,
      "epoch": 0.04,
      "learning_rate": 4.999705724966908e-05,
      "loss": 0.6819,
      "step": 39,
      "task_loss": 0.27272433042526245
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7013824687706396,
      "compression_loss": 0.0,
      "distillation_loss": 0.5879485011100769,
      "epoch": 0.04,
      "learning_rate": 4.999689149633402e-05,
      "loss": 0.5481,
      "step": 40,
      "task_loss": 0.18962328135967255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7014177481160824,
      "compression_loss": 0.0,
      "distillation_loss": 0.3988691568374634,
      "epoch": 0.04,
      "learning_rate": 4.999672120219323e-05,
      "loss": 0.3689,
      "step": 41,
      "task_loss": 0.09930090606212616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7014530190466545,
      "compression_loss": 0.0,
      "distillation_loss": 1.1346763372421265,
      "epoch": 0.04,
      "learning_rate": 4.999654636727764e-05,
      "loss": 1.0762,
      "step": 42,
      "task_loss": 0.5502893924713135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7014882815633595,
      "compression_loss": 0.0,
      "distillation_loss": 1.0310869216918945,
      "epoch": 0.04,
      "learning_rate": 4.9996366991619034e-05,
      "loss": 0.9707,
      "step": 43,
      "task_loss": 0.42731019854545593
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7015235356672014,
      "compression_loss": 0.0,
      "distillation_loss": 0.7622416615486145,
      "epoch": 0.04,
      "learning_rate": 4.999618307524997e-05,
      "loss": 0.7244,
      "step": 44,
      "task_loss": 0.38374945521354675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7015587813591835,
      "compression_loss": 0.0,
      "distillation_loss": 1.0716254711151123,
      "epoch": 0.04,
      "learning_rate": 4.999599461820387e-05,
      "loss": 1.0246,
      "step": 45,
      "task_loss": 0.6018234491348267
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7015940186403098,
      "compression_loss": 0.0,
      "distillation_loss": 0.7515419721603394,
      "epoch": 0.04,
      "learning_rate": 4.999580162051497e-05,
      "loss": 0.7087,
      "step": 46,
      "task_loss": 0.3226703405380249
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7016292475115837,
      "compression_loss": 0.0,
      "distillation_loss": 1.1621253490447998,
      "epoch": 0.04,
      "learning_rate": 4.9995604082218314e-05,
      "loss": 1.1132,
      "step": 47,
      "task_loss": 0.6729358434677124
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7016644679740092,
      "compression_loss": 0.0,
      "distillation_loss": 0.5425492525100708,
      "epoch": 0.05,
      "learning_rate": 4.99954020033498e-05,
      "loss": 0.5159,
      "step": 48,
      "task_loss": 0.2755749225616455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7016996800285896,
      "compression_loss": 0.0,
      "distillation_loss": 0.7733719348907471,
      "epoch": 0.05,
      "learning_rate": 4.9995195383946135e-05,
      "loss": 0.7346,
      "step": 49,
      "task_loss": 0.38580086827278137
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.701734883676329,
      "compression_loss": 0.0,
      "distillation_loss": 0.7129077911376953,
      "epoch": 0.05,
      "learning_rate": 4.999498422404485e-05,
      "loss": 0.6741,
      "step": 50,
      "task_loss": 0.3249415457248688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7017700789182307,
      "compression_loss": 0.0,
      "distillation_loss": 0.8956947922706604,
      "epoch": 0.05,
      "learning_rate": 4.999476852368431e-05,
      "loss": 0.8533,
      "step": 51,
      "task_loss": 0.4715935289859772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7018052657552987,
      "compression_loss": 0.0,
      "distillation_loss": 1.1765224933624268,
      "epoch": 0.05,
      "learning_rate": 4.999454828290369e-05,
      "loss": 1.1277,
      "step": 52,
      "task_loss": 0.6885514855384827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7018404441885364,
      "compression_loss": 0.0,
      "distillation_loss": 0.7294641137123108,
      "epoch": 0.05,
      "learning_rate": 4.999432350174299e-05,
      "loss": 0.6924,
      "step": 53,
      "task_loss": 0.3584412932395935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7018756142189477,
      "compression_loss": 0.0,
      "distillation_loss": 0.4732089936733246,
      "epoch": 0.05,
      "learning_rate": 4.9994094180243055e-05,
      "loss": 0.4407,
      "step": 54,
      "task_loss": 0.14808303117752075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7019107758475361,
      "compression_loss": 0.0,
      "distillation_loss": 0.7207993268966675,
      "epoch": 0.05,
      "learning_rate": 4.999386031844554e-05,
      "loss": 0.6882,
      "step": 55,
      "task_loss": 0.39489054679870605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7019459290753055,
      "compression_loss": 0.0,
      "distillation_loss": 0.7156393527984619,
      "epoch": 0.05,
      "learning_rate": 4.999362191639293e-05,
      "loss": 0.6781,
      "step": 56,
      "task_loss": 0.3404168486595154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7019810739032595,
      "compression_loss": 0.0,
      "distillation_loss": 0.528157651424408,
      "epoch": 0.05,
      "learning_rate": 4.999337897412852e-05,
      "loss": 0.4956,
      "step": 57,
      "task_loss": 0.20256038010120392
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7020162103324016,
      "compression_loss": 0.0,
      "distillation_loss": 0.7788621187210083,
      "epoch": 0.06,
      "learning_rate": 4.999313149169645e-05,
      "loss": 0.7403,
      "step": 58,
      "task_loss": 0.3931070566177368
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7020513383637357,
      "compression_loss": 0.0,
      "distillation_loss": 1.1968752145767212,
      "epoch": 0.06,
      "learning_rate": 4.999287946914169e-05,
      "loss": 1.1309,
      "step": 59,
      "task_loss": 0.5368590950965881
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7020864579982654,
      "compression_loss": 0.0,
      "distillation_loss": 0.7765418887138367,
      "epoch": 0.06,
      "learning_rate": 4.999262290651e-05,
      "loss": 0.7296,
      "step": 60,
      "task_loss": 0.3072131276130676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7021215692369944,
      "compression_loss": 0.0,
      "distillation_loss": 0.9318764209747314,
      "epoch": 0.06,
      "learning_rate": 4.9992361803847995e-05,
      "loss": 0.8886,
      "step": 61,
      "task_loss": 0.49911895394325256
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7021566720809264,
      "compression_loss": 0.0,
      "distillation_loss": 0.6415928602218628,
      "epoch": 0.06,
      "learning_rate": 4.99920961612031e-05,
      "loss": 0.6061,
      "step": 62,
      "task_loss": 0.2865428924560547
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.702191766531065,
      "compression_loss": 0.0,
      "distillation_loss": 0.8092690706253052,
      "epoch": 0.06,
      "learning_rate": 4.9991825978623574e-05,
      "loss": 0.7653,
      "step": 63,
      "task_loss": 0.3699354827404022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7022268525884139,
      "compression_loss": 0.0,
      "distillation_loss": 0.618144690990448,
      "epoch": 0.06,
      "learning_rate": 4.9991551256158495e-05,
      "loss": 0.5829,
      "step": 64,
      "task_loss": 0.26606857776641846
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7022619302539769,
      "compression_loss": 0.0,
      "distillation_loss": 0.8759365677833557,
      "epoch": 0.06,
      "learning_rate": 4.999127199385778e-05,
      "loss": 0.8231,
      "step": 65,
      "task_loss": 0.3478449881076813
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7022969995287576,
      "compression_loss": 0.0,
      "distillation_loss": 0.4816649556159973,
      "epoch": 0.06,
      "learning_rate": 4.999098819177214e-05,
      "loss": 0.4514,
      "step": 66,
      "task_loss": 0.17922161519527435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7023320604137597,
      "compression_loss": 0.0,
      "distillation_loss": 0.5684869289398193,
      "epoch": 0.06,
      "learning_rate": 4.999069984995314e-05,
      "loss": 0.535,
      "step": 67,
      "task_loss": 0.23375429213047028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7023671129099868,
      "compression_loss": 0.0,
      "distillation_loss": 0.7121763825416565,
      "epoch": 0.06,
      "learning_rate": 4.999040696845315e-05,
      "loss": 0.6668,
      "step": 68,
      "task_loss": 0.25801488757133484
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7024021570184427,
      "compression_loss": 0.0,
      "distillation_loss": 0.5653649568557739,
      "epoch": 0.07,
      "learning_rate": 4.999010954732538e-05,
      "loss": 0.5359,
      "step": 69,
      "task_loss": 0.27088284492492676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7024371927401311,
      "compression_loss": 0.0,
      "distillation_loss": 0.8046922087669373,
      "epoch": 0.07,
      "learning_rate": 4.998980758662386e-05,
      "loss": 0.7569,
      "step": 70,
      "task_loss": 0.3263823986053467
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7024722200760555,
      "compression_loss": 0.0,
      "distillation_loss": 0.6771596670150757,
      "epoch": 0.07,
      "learning_rate": 4.998950108640345e-05,
      "loss": 0.6458,
      "step": 71,
      "task_loss": 0.3636930286884308
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7025072390272197,
      "compression_loss": 0.0,
      "distillation_loss": 0.41592103242874146,
      "epoch": 0.07,
      "learning_rate": 4.99891900467198e-05,
      "loss": 0.3859,
      "step": 72,
      "task_loss": 0.11569308489561081
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7025422495946274,
      "compression_loss": 0.0,
      "distillation_loss": 0.5050232410430908,
      "epoch": 0.07,
      "learning_rate": 4.9988874467629435e-05,
      "loss": 0.4809,
      "step": 73,
      "task_loss": 0.26406604051589966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7025772517792822,
      "compression_loss": 0.0,
      "distillation_loss": 0.7662781476974487,
      "epoch": 0.07,
      "learning_rate": 4.998855434918968e-05,
      "loss": 0.7258,
      "step": 74,
      "task_loss": 0.36163684725761414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7026122455821879,
      "compression_loss": 0.0,
      "distillation_loss": 0.5249364376068115,
      "epoch": 0.07,
      "learning_rate": 4.998822969145868e-05,
      "loss": 0.4962,
      "step": 75,
      "task_loss": 0.23764313757419586
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7026472310043481,
      "compression_loss": 0.0,
      "distillation_loss": 0.755010187625885,
      "epoch": 0.07,
      "learning_rate": 4.99879004944954e-05,
      "loss": 0.7195,
      "step": 76,
      "task_loss": 0.4000910520553589
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7026822080467666,
      "compression_loss": 0.0,
      "distillation_loss": 0.5521838068962097,
      "epoch": 0.07,
      "learning_rate": 4.998756675835966e-05,
      "loss": 0.5156,
      "step": 77,
      "task_loss": 0.1858932077884674
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7027171767104469,
      "compression_loss": 0.0,
      "distillation_loss": 0.7275650501251221,
      "epoch": 0.07,
      "learning_rate": 4.9987228483112083e-05,
      "loss": 0.694,
      "step": 78,
      "task_loss": 0.39170515537261963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7027521369963928,
      "compression_loss": 0.0,
      "distillation_loss": 0.6621497869491577,
      "epoch": 0.08,
      "learning_rate": 4.998688566881411e-05,
      "loss": 0.6312,
      "step": 79,
      "task_loss": 0.35260745882987976
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7027870889056079,
      "compression_loss": 0.0,
      "distillation_loss": 0.8724344968795776,
      "epoch": 0.08,
      "learning_rate": 4.998653831552801e-05,
      "loss": 0.8453,
      "step": 80,
      "task_loss": 0.6009440422058105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.702822032439096,
      "compression_loss": 0.0,
      "distillation_loss": 0.6811240911483765,
      "epoch": 0.08,
      "learning_rate": 4.998618642331689e-05,
      "loss": 0.6469,
      "step": 81,
      "task_loss": 0.33839285373687744
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7028569675978606,
      "compression_loss": 0.0,
      "distillation_loss": 0.5384796857833862,
      "epoch": 0.08,
      "learning_rate": 4.9985829992244675e-05,
      "loss": 0.508,
      "step": 82,
      "task_loss": 0.2334153652191162
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7028918943829057,
      "compression_loss": 0.0,
      "distillation_loss": 0.8584867715835571,
      "epoch": 0.08,
      "learning_rate": 4.998546902237611e-05,
      "loss": 0.811,
      "step": 83,
      "task_loss": 0.38361459970474243
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7029268127952346,
      "compression_loss": 0.0,
      "distillation_loss": 0.419251024723053,
      "epoch": 0.08,
      "learning_rate": 4.9985103513776764e-05,
      "loss": 0.4111,
      "step": 84,
      "task_loss": 0.3377459645271301
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7029617228358512,
      "compression_loss": 0.0,
      "distillation_loss": 0.8462650179862976,
      "epoch": 0.08,
      "learning_rate": 4.998473346651303e-05,
      "loss": 0.8097,
      "step": 85,
      "task_loss": 0.48073211312294006
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7029966245057592,
      "compression_loss": 0.0,
      "distillation_loss": 1.191063404083252,
      "epoch": 0.08,
      "learning_rate": 4.9984358880652146e-05,
      "loss": 1.1373,
      "step": 86,
      "task_loss": 0.653830349445343
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7030315178059622,
      "compression_loss": 0.0,
      "distillation_loss": 0.8207377195358276,
      "epoch": 0.08,
      "learning_rate": 4.9983979756262136e-05,
      "loss": 0.7749,
      "step": 87,
      "task_loss": 0.3625227212905884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7030664027374639,
      "compression_loss": 0.0,
      "distillation_loss": 0.5961652994155884,
      "epoch": 0.08,
      "learning_rate": 4.998359609341188e-05,
      "loss": 0.5955,
      "step": 88,
      "task_loss": 0.589705228805542
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.703101279301268,
      "compression_loss": 0.0,
      "distillation_loss": 0.7842130661010742,
      "epoch": 0.08,
      "learning_rate": 4.9983207892171074e-05,
      "loss": 0.7404,
      "step": 89,
      "task_loss": 0.3462582230567932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7031361474983782,
      "compression_loss": 0.0,
      "distillation_loss": 0.8764528036117554,
      "epoch": 0.09,
      "learning_rate": 4.998281515261023e-05,
      "loss": 0.83,
      "step": 90,
      "task_loss": 0.41157200932502747
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7031710073297981,
      "compression_loss": 0.0,
      "distillation_loss": 1.0387656688690186,
      "epoch": 0.09,
      "learning_rate": 4.9982417874800704e-05,
      "loss": 0.9859,
      "step": 91,
      "task_loss": 0.5104459524154663
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7032058587965315,
      "compression_loss": 0.0,
      "distillation_loss": 0.4702117443084717,
      "epoch": 0.09,
      "learning_rate": 4.998201605881465e-05,
      "loss": 0.4451,
      "step": 92,
      "task_loss": 0.2186840921640396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.703240701899582,
      "compression_loss": 0.0,
      "distillation_loss": 0.5508553981781006,
      "epoch": 0.09,
      "learning_rate": 4.9981609704725057e-05,
      "loss": 0.516,
      "step": 93,
      "task_loss": 0.20186278223991394
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7032755366399532,
      "compression_loss": 0.0,
      "distillation_loss": 0.4743342697620392,
      "epoch": 0.09,
      "learning_rate": 4.998119881260576e-05,
      "loss": 0.4436,
      "step": 94,
      "task_loss": 0.16690446436405182
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.703310363018649,
      "compression_loss": 0.0,
      "distillation_loss": 0.7569644451141357,
      "epoch": 0.09,
      "learning_rate": 4.9980783382531376e-05,
      "loss": 0.7158,
      "step": 95,
      "task_loss": 0.3451739251613617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7033451810366729,
      "compression_loss": 0.0,
      "distillation_loss": 0.6965117454528809,
      "epoch": 0.09,
      "learning_rate": 4.998036341457739e-05,
      "loss": 0.659,
      "step": 96,
      "task_loss": 0.32098302245140076
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7033799906950287,
      "compression_loss": 0.0,
      "distillation_loss": 0.8976222276687622,
      "epoch": 0.09,
      "learning_rate": 4.997993890882008e-05,
      "loss": 0.8545,
      "step": 97,
      "task_loss": 0.46677258610725403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.70341479199472,
      "compression_loss": 0.0,
      "distillation_loss": 1.0592153072357178,
      "epoch": 0.09,
      "learning_rate": 4.997950986533656e-05,
      "loss": 1.0024,
      "step": 98,
      "task_loss": 0.4912152886390686
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7034495849367505,
      "compression_loss": 0.0,
      "distillation_loss": 0.7887797355651855,
      "epoch": 0.09,
      "learning_rate": 4.997907628420477e-05,
      "loss": 0.7517,
      "step": 99,
      "task_loss": 0.4175463318824768
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7034843695221239,
      "compression_loss": 0.0,
      "distillation_loss": 0.7304688692092896,
      "epoch": 0.09,
      "learning_rate": 4.9978638165503475e-05,
      "loss": 0.6888,
      "step": 100,
      "task_loss": 0.3141331076622009
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7035191457518439,
      "compression_loss": 0.0,
      "distillation_loss": 0.7318391799926758,
      "epoch": 0.1,
      "learning_rate": 4.9978195509312266e-05,
      "loss": 0.7035,
      "step": 101,
      "task_loss": 0.4489053189754486
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7035539136269141,
      "compression_loss": 0.0,
      "distillation_loss": 0.5332415699958801,
      "epoch": 0.1,
      "learning_rate": 4.997774831571154e-05,
      "loss": 0.5093,
      "step": 102,
      "task_loss": 0.2942197620868683
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7035886731483383,
      "compression_loss": 0.0,
      "distillation_loss": 0.6230442523956299,
      "epoch": 0.1,
      "learning_rate": 4.9977296584782544e-05,
      "loss": 0.5832,
      "step": 103,
      "task_loss": 0.2250797152519226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7036234243171201,
      "compression_loss": 0.0,
      "distillation_loss": 0.28762802481651306,
      "epoch": 0.1,
      "learning_rate": 4.997684031660732e-05,
      "loss": 0.2797,
      "step": 104,
      "task_loss": 0.20829369127750397
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7036581671342632,
      "compression_loss": 0.0,
      "distillation_loss": 0.7059702277183533,
      "epoch": 0.1,
      "learning_rate": 4.997637951126877e-05,
      "loss": 0.6714,
      "step": 105,
      "task_loss": 0.36017781496047974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7036929016007714,
      "compression_loss": 0.0,
      "distillation_loss": 0.9591817855834961,
      "epoch": 0.1,
      "learning_rate": 4.997591416885059e-05,
      "loss": 0.9062,
      "step": 106,
      "task_loss": 0.4291348457336426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7037276277176481,
      "compression_loss": 0.0,
      "distillation_loss": 0.5214189887046814,
      "epoch": 0.1,
      "learning_rate": 4.997544428943732e-05,
      "loss": 0.4899,
      "step": 107,
      "task_loss": 0.20646995306015015
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7037623454858972,
      "compression_loss": 0.0,
      "distillation_loss": 0.7594773769378662,
      "epoch": 0.1,
      "learning_rate": 4.997496987311431e-05,
      "loss": 0.7246,
      "step": 108,
      "task_loss": 0.41065266728401184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7037970549065223,
      "compression_loss": 0.0,
      "distillation_loss": 1.0592186450958252,
      "epoch": 0.1,
      "learning_rate": 4.997449091996774e-05,
      "loss": 1.0102,
      "step": 109,
      "task_loss": 0.5689947605133057
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7038317559805272,
      "compression_loss": 0.0,
      "distillation_loss": 0.717729389667511,
      "epoch": 0.1,
      "learning_rate": 4.9974007430084617e-05,
      "loss": 0.693,
      "step": 110,
      "task_loss": 0.47034698724746704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7038664487089155,
      "compression_loss": 0.0,
      "distillation_loss": 0.5633364915847778,
      "epoch": 0.11,
      "learning_rate": 4.997351940355277e-05,
      "loss": 0.5359,
      "step": 111,
      "task_loss": 0.2890656292438507
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7039011330926908,
      "compression_loss": 0.0,
      "distillation_loss": 0.8346014618873596,
      "epoch": 0.11,
      "learning_rate": 4.997302684046085e-05,
      "loss": 0.7913,
      "step": 112,
      "task_loss": 0.4014682173728943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7039358091328568,
      "compression_loss": 0.0,
      "distillation_loss": 0.6383477449417114,
      "epoch": 0.11,
      "learning_rate": 4.997252974089833e-05,
      "loss": 0.6,
      "step": 113,
      "task_loss": 0.25466716289520264
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7039704768304174,
      "compression_loss": 0.0,
      "distillation_loss": 1.142728328704834,
      "epoch": 0.11,
      "learning_rate": 4.997202810495551e-05,
      "loss": 1.0939,
      "step": 114,
      "task_loss": 0.6547843813896179
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7040051361863761,
      "compression_loss": 0.0,
      "distillation_loss": 0.4164957106113434,
      "epoch": 0.11,
      "learning_rate": 4.997152193272353e-05,
      "loss": 0.391,
      "step": 115,
      "task_loss": 0.16178376972675323
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7040397872017365,
      "compression_loss": 0.0,
      "distillation_loss": 0.24880273640155792,
      "epoch": 0.11,
      "learning_rate": 4.9971011224294314e-05,
      "loss": 0.2379,
      "step": 116,
      "task_loss": 0.14008797705173492
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7040744298775025,
      "compression_loss": 0.0,
      "distillation_loss": 0.9239636659622192,
      "epoch": 0.11,
      "learning_rate": 4.997049597976066e-05,
      "loss": 0.884,
      "step": 117,
      "task_loss": 0.5244091749191284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7041090642146776,
      "compression_loss": 0.0,
      "distillation_loss": 0.43864983320236206,
      "epoch": 0.11,
      "learning_rate": 4.9969976199216144e-05,
      "loss": 0.4116,
      "step": 118,
      "task_loss": 0.16765302419662476
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7041436902142656,
      "compression_loss": 0.0,
      "distillation_loss": 0.5404030084609985,
      "epoch": 0.11,
      "learning_rate": 4.9969451882755196e-05,
      "loss": 0.5152,
      "step": 119,
      "task_loss": 0.28787752985954285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7041783078772701,
      "compression_loss": 0.0,
      "distillation_loss": 0.4323447644710541,
      "epoch": 0.11,
      "learning_rate": 4.996892303047306e-05,
      "loss": 0.4133,
      "step": 120,
      "task_loss": 0.24157865345478058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7042129172046948,
      "compression_loss": 0.0,
      "distillation_loss": 0.6243703365325928,
      "epoch": 0.11,
      "learning_rate": 4.996838964246581e-05,
      "loss": 0.5941,
      "step": 121,
      "task_loss": 0.3215750753879547
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7042475181975434,
      "compression_loss": 0.0,
      "distillation_loss": 0.7456376552581787,
      "epoch": 0.12,
      "learning_rate": 4.996785171883032e-05,
      "loss": 0.7073,
      "step": 122,
      "task_loss": 0.3622223436832428
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7042821108568197,
      "compression_loss": 0.0,
      "distillation_loss": 0.7294185161590576,
      "epoch": 0.12,
      "learning_rate": 4.996730925966433e-05,
      "loss": 0.694,
      "step": 123,
      "task_loss": 0.3748391270637512
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7043166951835271,
      "compression_loss": 0.0,
      "distillation_loss": 1.1894690990447998,
      "epoch": 0.12,
      "learning_rate": 4.996676226506636e-05,
      "loss": 1.1356,
      "step": 124,
      "task_loss": 0.6504833102226257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7043512711786696,
      "compression_loss": 0.0,
      "distillation_loss": 0.6352778673171997,
      "epoch": 0.12,
      "learning_rate": 4.9966210735135785e-05,
      "loss": 0.6021,
      "step": 125,
      "task_loss": 0.3033022880554199
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7043858388432506,
      "compression_loss": 0.0,
      "distillation_loss": 0.4930354058742523,
      "epoch": 0.12,
      "learning_rate": 4.9965654669972794e-05,
      "loss": 0.4721,
      "step": 126,
      "task_loss": 0.2833815813064575
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.704420398178274,
      "compression_loss": 0.0,
      "distillation_loss": 1.0228924751281738,
      "epoch": 0.12,
      "learning_rate": 4.99650940696784e-05,
      "loss": 0.9778,
      "step": 127,
      "task_loss": 0.571823000907898
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7044549491847434,
      "compression_loss": 0.0,
      "distillation_loss": 0.5655504465103149,
      "epoch": 0.12,
      "learning_rate": 4.996452893435442e-05,
      "loss": 0.5301,
      "step": 128,
      "task_loss": 0.21055805683135986
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7044894918636624,
      "compression_loss": 0.0,
      "distillation_loss": 0.6770893335342407,
      "epoch": 0.12,
      "learning_rate": 4.9963959264103544e-05,
      "loss": 0.6397,
      "step": 129,
      "task_loss": 0.30343693494796753
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7045240262160347,
      "compression_loss": 0.0,
      "distillation_loss": 0.4913747012615204,
      "epoch": 0.12,
      "learning_rate": 4.996338505902924e-05,
      "loss": 0.4605,
      "step": 130,
      "task_loss": 0.1823093295097351
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7045585522428641,
      "compression_loss": 0.0,
      "distillation_loss": 0.5457890033721924,
      "epoch": 0.12,
      "learning_rate": 4.996280631923581e-05,
      "loss": 0.5186,
      "step": 131,
      "task_loss": 0.2735254168510437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7045930699451544,
      "compression_loss": 0.0,
      "distillation_loss": 0.2785548269748688,
      "epoch": 0.13,
      "learning_rate": 4.9962223044828396e-05,
      "loss": 0.2587,
      "step": 132,
      "task_loss": 0.08010871708393097
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7046275793239088,
      "compression_loss": 0.0,
      "distillation_loss": 0.49103331565856934,
      "epoch": 0.13,
      "learning_rate": 4.9961635235912935e-05,
      "loss": 0.4746,
      "step": 133,
      "task_loss": 0.32637351751327515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7046620803801315,
      "compression_loss": 0.0,
      "distillation_loss": 0.3778892755508423,
      "epoch": 0.13,
      "learning_rate": 4.9961042892596225e-05,
      "loss": 0.3622,
      "step": 134,
      "task_loss": 0.2211739420890808
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7046965731148259,
      "compression_loss": 0.0,
      "distillation_loss": 0.4187111258506775,
      "epoch": 0.13,
      "learning_rate": 4.996044601498586e-05,
      "loss": 0.3963,
      "step": 135,
      "task_loss": 0.1942349374294281
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7047310575289957,
      "compression_loss": 0.0,
      "distillation_loss": 0.5608755350112915,
      "epoch": 0.13,
      "learning_rate": 4.995984460319026e-05,
      "loss": 0.5274,
      "step": 136,
      "task_loss": 0.22652751207351685
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7047655336236447,
      "compression_loss": 0.0,
      "distillation_loss": 0.642856776714325,
      "epoch": 0.13,
      "learning_rate": 4.995923865731869e-05,
      "loss": 0.6072,
      "step": 137,
      "task_loss": 0.28608477115631104
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7048000013997765,
      "compression_loss": 0.0,
      "distillation_loss": 0.6759918928146362,
      "epoch": 0.13,
      "learning_rate": 4.9958628177481195e-05,
      "loss": 0.6456,
      "step": 138,
      "task_loss": 0.37211018800735474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7048344608583947,
      "compression_loss": 0.0,
      "distillation_loss": 0.5849887728691101,
      "epoch": 0.13,
      "learning_rate": 4.99580131637887e-05,
      "loss": 0.5497,
      "step": 139,
      "task_loss": 0.23251289129257202
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7048689120005032,
      "compression_loss": 0.0,
      "distillation_loss": 0.5638033151626587,
      "epoch": 0.13,
      "learning_rate": 4.995739361635292e-05,
      "loss": 0.5441,
      "step": 140,
      "task_loss": 0.367082804441452
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7049033548271055,
      "compression_loss": 0.0,
      "distillation_loss": 0.4791497588157654,
      "epoch": 0.13,
      "learning_rate": 4.9956769535286385e-05,
      "loss": 0.4515,
      "step": 141,
      "task_loss": 0.20264172554016113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7049377893392055,
      "compression_loss": 0.0,
      "distillation_loss": 0.45656701922416687,
      "epoch": 0.13,
      "learning_rate": 4.9956140920702476e-05,
      "loss": 0.4341,
      "step": 142,
      "task_loss": 0.23189952969551086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7049722155378065,
      "compression_loss": 0.0,
      "distillation_loss": 0.5628130435943604,
      "epoch": 0.14,
      "learning_rate": 4.995550777271538e-05,
      "loss": 0.5244,
      "step": 143,
      "task_loss": 0.1786140650510788
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7050066334239125,
      "compression_loss": 0.0,
      "distillation_loss": 0.8593398332595825,
      "epoch": 0.14,
      "learning_rate": 4.995487009144011e-05,
      "loss": 0.817,
      "step": 144,
      "task_loss": 0.43626001477241516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7050410429985271,
      "compression_loss": 0.0,
      "distillation_loss": 0.6192554235458374,
      "epoch": 0.14,
      "learning_rate": 4.99542278769925e-05,
      "loss": 0.5859,
      "step": 145,
      "task_loss": 0.2856113016605377
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7050754442626539,
      "compression_loss": 0.0,
      "distillation_loss": 0.46989959478378296,
      "epoch": 0.14,
      "learning_rate": 4.995358112948921e-05,
      "loss": 0.4416,
      "step": 146,
      "task_loss": 0.18703171610832214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7051098372172967,
      "compression_loss": 0.0,
      "distillation_loss": 0.7966834902763367,
      "epoch": 0.14,
      "learning_rate": 4.9952929849047734e-05,
      "loss": 0.754,
      "step": 147,
      "task_loss": 0.3696812093257904
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7051442218634592,
      "compression_loss": 0.0,
      "distillation_loss": 0.8801038265228271,
      "epoch": 0.14,
      "learning_rate": 4.9952274035786385e-05,
      "loss": 0.8282,
      "step": 148,
      "task_loss": 0.3614765405654907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7051785982021449,
      "compression_loss": 0.0,
      "distillation_loss": 0.8862295150756836,
      "epoch": 0.14,
      "learning_rate": 4.9951613689824276e-05,
      "loss": 0.8481,
      "step": 149,
      "task_loss": 0.5047279596328735
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7052129662343577,
      "compression_loss": 0.0,
      "distillation_loss": 0.7615218162536621,
      "epoch": 0.14,
      "learning_rate": 4.995094881128138e-05,
      "loss": 0.7294,
      "step": 150,
      "task_loss": 0.4406706392765045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7052473259611011,
      "compression_loss": 0.0,
      "distillation_loss": 0.5523242950439453,
      "epoch": 0.14,
      "learning_rate": 4.995027940027846e-05,
      "loss": 0.5313,
      "step": 151,
      "task_loss": 0.3420305848121643
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7052816773833789,
      "compression_loss": 0.0,
      "distillation_loss": 0.43813109397888184,
      "epoch": 0.14,
      "learning_rate": 4.9949605456937135e-05,
      "loss": 0.4062,
      "step": 152,
      "task_loss": 0.11931411176919937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7053160205021947,
      "compression_loss": 0.0,
      "distillation_loss": 0.6433530449867249,
      "epoch": 0.15,
      "learning_rate": 4.994892698137981e-05,
      "loss": 0.608,
      "step": 153,
      "task_loss": 0.2893901765346527
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7053503553185523,
      "compression_loss": 0.0,
      "distillation_loss": 0.6064528226852417,
      "epoch": 0.15,
      "learning_rate": 4.9948243973729745e-05,
      "loss": 0.5714,
      "step": 154,
      "task_loss": 0.25594377517700195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7053846818334553,
      "compression_loss": 0.0,
      "distillation_loss": 0.7984007000923157,
      "epoch": 0.15,
      "learning_rate": 4.994755643411101e-05,
      "loss": 0.7624,
      "step": 155,
      "task_loss": 0.43833673000335693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7054190000479073,
      "compression_loss": 0.0,
      "distillation_loss": 0.5333908796310425,
      "epoch": 0.15,
      "learning_rate": 4.9946864362648506e-05,
      "loss": 0.5041,
      "step": 156,
      "task_loss": 0.240193709731102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7054533099629121,
      "compression_loss": 0.0,
      "distillation_loss": 0.5313940048217773,
      "epoch": 0.15,
      "learning_rate": 4.994616775946794e-05,
      "loss": 0.5092,
      "step": 157,
      "task_loss": 0.3092755973339081
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7054876115794735,
      "compression_loss": 0.0,
      "distillation_loss": 0.6694386005401611,
      "epoch": 0.15,
      "learning_rate": 4.994546662469586e-05,
      "loss": 0.6325,
      "step": 158,
      "task_loss": 0.3002847731113434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7055219048985948,
      "compression_loss": 0.0,
      "distillation_loss": 0.43738991022109985,
      "epoch": 0.15,
      "learning_rate": 4.9944760958459624e-05,
      "loss": 0.4088,
      "step": 159,
      "task_loss": 0.15195739269256592
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7055561899212801,
      "compression_loss": 0.0,
      "distillation_loss": 0.6324198842048645,
      "epoch": 0.15,
      "learning_rate": 4.994405076088743e-05,
      "loss": 0.6241,
      "step": 160,
      "task_loss": 0.5489103198051453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7055904666485329,
      "compression_loss": 0.0,
      "distillation_loss": 0.4191284775733948,
      "epoch": 0.15,
      "learning_rate": 4.994333603210829e-05,
      "loss": 0.4006,
      "step": 161,
      "task_loss": 0.23342673480510712
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7056247350813567,
      "compression_loss": 0.0,
      "distillation_loss": 0.5982824563980103,
      "epoch": 0.15,
      "learning_rate": 4.9942616772252016e-05,
      "loss": 0.5746,
      "step": 162,
      "task_loss": 0.3616164028644562
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7056589952207556,
      "compression_loss": 0.0,
      "distillation_loss": 0.279816210269928,
      "epoch": 0.15,
      "learning_rate": 4.994189298144929e-05,
      "loss": 0.2647,
      "step": 163,
      "task_loss": 0.12842896580696106
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7056932470677328,
      "compression_loss": 0.0,
      "distillation_loss": 0.26615631580352783,
      "epoch": 0.16,
      "learning_rate": 4.994116465983158e-05,
      "loss": 0.2481,
      "step": 164,
      "task_loss": 0.08531204611063004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7057274906232924,
      "compression_loss": 0.0,
      "distillation_loss": 0.7699984908103943,
      "epoch": 0.16,
      "learning_rate": 4.99404318075312e-05,
      "loss": 0.7333,
      "step": 165,
      "task_loss": 0.4031785726547241
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7057617258884379,
      "compression_loss": 0.0,
      "distillation_loss": 0.32784250378608704,
      "epoch": 0.16,
      "learning_rate": 4.993969442468125e-05,
      "loss": 0.3131,
      "step": 166,
      "task_loss": 0.1799994260072708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7057959528641731,
      "compression_loss": 0.0,
      "distillation_loss": 0.5546020269393921,
      "epoch": 0.16,
      "learning_rate": 4.993895251141571e-05,
      "loss": 0.5365,
      "step": 167,
      "task_loss": 0.37407755851745605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7058301715515014,
      "compression_loss": 0.0,
      "distillation_loss": 0.24405278265476227,
      "epoch": 0.16,
      "learning_rate": 4.9938206067869334e-05,
      "loss": 0.2312,
      "step": 168,
      "task_loss": 0.11586499214172363
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7058643819514266,
      "compression_loss": 0.0,
      "distillation_loss": 0.897857666015625,
      "epoch": 0.16,
      "learning_rate": 4.993745509417772e-05,
      "loss": 0.8501,
      "step": 169,
      "task_loss": 0.42022186517715454
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7058985840649525,
      "compression_loss": 0.0,
      "distillation_loss": 0.62211012840271,
      "epoch": 0.16,
      "learning_rate": 4.9936699590477296e-05,
      "loss": 0.5941,
      "step": 170,
      "task_loss": 0.3424464166164398
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7059327778930828,
      "compression_loss": 0.0,
      "distillation_loss": 0.4366634488105774,
      "epoch": 0.16,
      "learning_rate": 4.9935939556905295e-05,
      "loss": 0.4205,
      "step": 171,
      "task_loss": 0.27515465021133423
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7059669634368212,
      "compression_loss": 0.0,
      "distillation_loss": 0.4915463626384735,
      "epoch": 0.16,
      "learning_rate": 4.993517499359978e-05,
      "loss": 0.4641,
      "step": 172,
      "task_loss": 0.2167641520500183
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7060011406971711,
      "compression_loss": 0.0,
      "distillation_loss": 0.5092770457267761,
      "epoch": 0.16,
      "learning_rate": 4.993440590069963e-05,
      "loss": 0.4841,
      "step": 173,
      "task_loss": 0.257138729095459
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7060353096751364,
      "compression_loss": 0.0,
      "distillation_loss": 0.5051209926605225,
      "epoch": 0.17,
      "learning_rate": 4.993363227834457e-05,
      "loss": 0.4766,
      "step": 174,
      "task_loss": 0.22033719718456268
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7060694703717209,
      "compression_loss": 0.0,
      "distillation_loss": 0.6676339507102966,
      "epoch": 0.17,
      "learning_rate": 4.9932854126675124e-05,
      "loss": 0.6391,
      "step": 175,
      "task_loss": 0.3822685778141022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7061036227879279,
      "compression_loss": 0.0,
      "distillation_loss": 0.2579382061958313,
      "epoch": 0.17,
      "learning_rate": 4.993207144583264e-05,
      "loss": 0.2493,
      "step": 176,
      "task_loss": 0.17196613550186157
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7061377669247615,
      "compression_loss": 0.0,
      "distillation_loss": 0.39016398787498474,
      "epoch": 0.17,
      "learning_rate": 4.993128423595931e-05,
      "loss": 0.3674,
      "step": 177,
      "task_loss": 0.16221883893013
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7061719027832251,
      "compression_loss": 0.0,
      "distillation_loss": 0.5886770486831665,
      "epoch": 0.17,
      "learning_rate": 4.9930492497198125e-05,
      "loss": 0.5569,
      "step": 178,
      "task_loss": 0.2707720994949341
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7062060303643225,
      "compression_loss": 0.0,
      "distillation_loss": 0.31564319133758545,
      "epoch": 0.17,
      "learning_rate": 4.992969622969292e-05,
      "loss": 0.3039,
      "step": 179,
      "task_loss": 0.19809630513191223
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7062401496690573,
      "compression_loss": 0.0,
      "distillation_loss": 0.35320743918418884,
      "epoch": 0.17,
      "learning_rate": 4.992889543358832e-05,
      "loss": 0.3371,
      "step": 180,
      "task_loss": 0.19164326786994934
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7062742606984334,
      "compression_loss": 0.0,
      "distillation_loss": 0.8013370633125305,
      "epoch": 0.17,
      "learning_rate": 4.9928090109029817e-05,
      "loss": 0.7688,
      "step": 181,
      "task_loss": 0.4764384627342224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7063083634534542,
      "compression_loss": 0.0,
      "distillation_loss": 1.4201560020446777,
      "epoch": 0.17,
      "learning_rate": 4.9927280256163686e-05,
      "loss": 1.3439,
      "step": 182,
      "task_loss": 0.6578447222709656
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7063424579351235,
      "compression_loss": 0.0,
      "distillation_loss": 0.35098546743392944,
      "epoch": 0.17,
      "learning_rate": 4.992646587513705e-05,
      "loss": 0.3229,
      "step": 183,
      "task_loss": 0.07008456438779831
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7063765441444451,
      "compression_loss": 0.0,
      "distillation_loss": 0.6404779553413391,
      "epoch": 0.17,
      "learning_rate": 4.9925646966097835e-05,
      "loss": 0.6098,
      "step": 184,
      "task_loss": 0.3341161012649536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7064106220824224,
      "compression_loss": 0.0,
      "distillation_loss": 0.4935486912727356,
      "epoch": 0.18,
      "learning_rate": 4.99248235291948e-05,
      "loss": 0.4679,
      "step": 185,
      "task_loss": 0.2367597073316574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7064446917500594,
      "compression_loss": 0.0,
      "distillation_loss": 0.3535561263561249,
      "epoch": 0.18,
      "learning_rate": 4.9923995564577544e-05,
      "loss": 0.3314,
      "step": 186,
      "task_loss": 0.13228961825370789
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7064787531483596,
      "compression_loss": 0.0,
      "distillation_loss": 0.6032556295394897,
      "epoch": 0.18,
      "learning_rate": 4.992316307239645e-05,
      "loss": 0.572,
      "step": 187,
      "task_loss": 0.2907094657421112
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7065128062783267,
      "compression_loss": 0.0,
      "distillation_loss": 1.0592761039733887,
      "epoch": 0.18,
      "learning_rate": 4.992232605280276e-05,
      "loss": 1.0138,
      "step": 188,
      "task_loss": 0.6046409606933594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7065468511409644,
      "compression_loss": 0.0,
      "distillation_loss": 0.32051318883895874,
      "epoch": 0.18,
      "learning_rate": 4.992148450594851e-05,
      "loss": 0.3152,
      "step": 189,
      "task_loss": 0.26700055599212646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7065808877372763,
      "compression_loss": 0.0,
      "distillation_loss": 0.4926970899105072,
      "epoch": 0.18,
      "learning_rate": 4.9920638431986574e-05,
      "loss": 0.4646,
      "step": 190,
      "task_loss": 0.21135534346103668
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7066149160682663,
      "compression_loss": 0.0,
      "distillation_loss": 0.7110521793365479,
      "epoch": 0.18,
      "learning_rate": 4.991978783107065e-05,
      "loss": 0.6783,
      "step": 191,
      "task_loss": 0.38350385427474976
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.706648936134938,
      "compression_loss": 0.0,
      "distillation_loss": 0.4119304120540619,
      "epoch": 0.18,
      "learning_rate": 4.9918932703355256e-05,
      "loss": 0.3859,
      "step": 192,
      "task_loss": 0.15174490213394165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7066829479382948,
      "compression_loss": 0.0,
      "distillation_loss": 0.5923248529434204,
      "epoch": 0.18,
      "learning_rate": 4.991807304899572e-05,
      "loss": 0.5592,
      "step": 193,
      "task_loss": 0.26117557287216187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7067169514793408,
      "compression_loss": 0.0,
      "distillation_loss": 0.41325411200523376,
      "epoch": 0.18,
      "learning_rate": 4.991720886814821e-05,
      "loss": 0.3904,
      "step": 194,
      "task_loss": 0.18504031002521515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7067509467590795,
      "compression_loss": 0.0,
      "distillation_loss": 0.4917716085910797,
      "epoch": 0.19,
      "learning_rate": 4.99163401609697e-05,
      "loss": 0.4618,
      "step": 195,
      "task_loss": 0.1917870044708252
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7067849337785144,
      "compression_loss": 0.0,
      "distillation_loss": 0.48008596897125244,
      "epoch": 0.19,
      "learning_rate": 4.991546692761801e-05,
      "loss": 0.4578,
      "step": 196,
      "task_loss": 0.25695863366127014
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7068189125386495,
      "compression_loss": 0.0,
      "distillation_loss": 0.8246874213218689,
      "epoch": 0.19,
      "learning_rate": 4.991458916825176e-05,
      "loss": 0.7932,
      "step": 197,
      "task_loss": 0.5102267265319824
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7068528830404883,
      "compression_loss": 0.0,
      "distillation_loss": 0.49332180619239807,
      "epoch": 0.19,
      "learning_rate": 4.991370688303039e-05,
      "loss": 0.4696,
      "step": 198,
      "task_loss": 0.25608059763908386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7068868452850345,
      "compression_loss": 0.0,
      "distillation_loss": 0.7726448178291321,
      "epoch": 0.19,
      "learning_rate": 4.9912820072114185e-05,
      "loss": 0.7387,
      "step": 199,
      "task_loss": 0.43347033858299255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7069207992732918,
      "compression_loss": 0.0,
      "distillation_loss": 0.3915247917175293,
      "epoch": 0.19,
      "learning_rate": 4.9911928735664224e-05,
      "loss": 0.368,
      "step": 200,
      "task_loss": 0.1565304845571518
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7069547450062639,
      "compression_loss": 0.0,
      "distillation_loss": 0.39359933137893677,
      "epoch": 0.19,
      "learning_rate": 4.991103287384244e-05,
      "loss": 0.3707,
      "step": 201,
      "task_loss": 0.1646425426006317
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7069886824849545,
      "compression_loss": 0.0,
      "distillation_loss": 0.30474036931991577,
      "epoch": 0.19,
      "learning_rate": 4.9910132486811555e-05,
      "loss": 0.2895,
      "step": 202,
      "task_loss": 0.15248741209506989
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7070226117103672,
      "compression_loss": 0.0,
      "distillation_loss": 0.5055716037750244,
      "epoch": 0.19,
      "learning_rate": 4.990922757473514e-05,
      "loss": 0.4763,
      "step": 203,
      "task_loss": 0.21311217546463013
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7070565326835058,
      "compression_loss": 0.0,
      "distillation_loss": 0.6619958877563477,
      "epoch": 0.19,
      "learning_rate": 4.990831813777757e-05,
      "loss": 0.6328,
      "step": 204,
      "task_loss": 0.36974024772644043
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7070904454053738,
      "compression_loss": 0.0,
      "distillation_loss": 0.809003472328186,
      "epoch": 0.19,
      "learning_rate": 4.990740417610406e-05,
      "loss": 0.7783,
      "step": 205,
      "task_loss": 0.5022717714309692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.707124349876975,
      "compression_loss": 0.0,
      "distillation_loss": 0.9417250156402588,
      "epoch": 0.2,
      "learning_rate": 4.9906485689880613e-05,
      "loss": 0.898,
      "step": 206,
      "task_loss": 0.5048108100891113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7071582460993132,
      "compression_loss": 0.0,
      "distillation_loss": 0.4026396870613098,
      "epoch": 0.2,
      "learning_rate": 4.9905562679274096e-05,
      "loss": 0.3856,
      "step": 207,
      "task_loss": 0.23229239881038666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7071921340733919,
      "compression_loss": 0.0,
      "distillation_loss": 0.45894497632980347,
      "epoch": 0.2,
      "learning_rate": 4.9904635144452164e-05,
      "loss": 0.4355,
      "step": 208,
      "task_loss": 0.22411760687828064
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7072260138002149,
      "compression_loss": 0.0,
      "distillation_loss": 0.5892167091369629,
      "epoch": 0.2,
      "learning_rate": 4.990370308558332e-05,
      "loss": 0.5551,
      "step": 209,
      "task_loss": 0.24835461378097534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7072598852807856,
      "compression_loss": 0.0,
      "distillation_loss": 0.5774100422859192,
      "epoch": 0.2,
      "learning_rate": 4.9902766502836874e-05,
      "loss": 0.5463,
      "step": 210,
      "task_loss": 0.2667173445224762
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7072937485161082,
      "compression_loss": 0.0,
      "distillation_loss": 0.33134809136390686,
      "epoch": 0.2,
      "learning_rate": 4.9901825396382965e-05,
      "loss": 0.3096,
      "step": 211,
      "task_loss": 0.11349479109048843
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7073276035071859,
      "compression_loss": 0.0,
      "distillation_loss": 0.46322399377822876,
      "epoch": 0.2,
      "learning_rate": 4.990087976639254e-05,
      "loss": 0.4346,
      "step": 212,
      "task_loss": 0.17661762237548828
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7073614502550227,
      "compression_loss": 0.0,
      "distillation_loss": 0.4679402709007263,
      "epoch": 0.2,
      "learning_rate": 4.989992961303738e-05,
      "loss": 0.445,
      "step": 213,
      "task_loss": 0.23898877203464508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7073952887606221,
      "compression_loss": 0.0,
      "distillation_loss": 0.44021981954574585,
      "epoch": 0.2,
      "learning_rate": 4.989897493649008e-05,
      "loss": 0.43,
      "step": 214,
      "task_loss": 0.3383277654647827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7074291190249877,
      "compression_loss": 0.0,
      "distillation_loss": 0.5056917071342468,
      "epoch": 0.2,
      "learning_rate": 4.989801573692408e-05,
      "loss": 0.4787,
      "step": 215,
      "task_loss": 0.23537424206733704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7074629410491236,
      "compression_loss": 0.0,
      "distillation_loss": 0.7158606052398682,
      "epoch": 0.21,
      "learning_rate": 4.989705201451361e-05,
      "loss": 0.6868,
      "step": 216,
      "task_loss": 0.4253765344619751
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.707496754834033,
      "compression_loss": 0.0,
      "distillation_loss": 0.4985395669937134,
      "epoch": 0.21,
      "learning_rate": 4.989608376943373e-05,
      "loss": 0.4697,
      "step": 217,
      "task_loss": 0.21007469296455383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7075305603807197,
      "compression_loss": 0.0,
      "distillation_loss": 0.5073490738868713,
      "epoch": 0.21,
      "learning_rate": 4.9895111001860335e-05,
      "loss": 0.4779,
      "step": 218,
      "task_loss": 0.21258264780044556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7075643576901876,
      "compression_loss": 0.0,
      "distillation_loss": 0.37614893913269043,
      "epoch": 0.21,
      "learning_rate": 4.989413371197013e-05,
      "loss": 0.3534,
      "step": 219,
      "task_loss": 0.14843641221523285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7075981467634402,
      "compression_loss": 0.0,
      "distillation_loss": 0.6342090368270874,
      "epoch": 0.21,
      "learning_rate": 4.989315189994065e-05,
      "loss": 0.6009,
      "step": 220,
      "task_loss": 0.3015548586845398
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7076319276014813,
      "compression_loss": 0.0,
      "distillation_loss": 0.7175476551055908,
      "epoch": 0.21,
      "learning_rate": 4.9892165565950235e-05,
      "loss": 0.6756,
      "step": 221,
      "task_loss": 0.2982741594314575
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7076657002053144,
      "compression_loss": 0.0,
      "distillation_loss": 0.40784966945648193,
      "epoch": 0.21,
      "learning_rate": 4.9891174710178054e-05,
      "loss": 0.389,
      "step": 222,
      "task_loss": 0.21939508616924286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7076994645759433,
      "compression_loss": 0.0,
      "distillation_loss": 0.5739084482192993,
      "epoch": 0.21,
      "learning_rate": 4.9890179332804125e-05,
      "loss": 0.5553,
      "step": 223,
      "task_loss": 0.38766252994537354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7077332207143717,
      "compression_loss": 0.0,
      "distillation_loss": 0.44305968284606934,
      "epoch": 0.21,
      "learning_rate": 4.988917943400924e-05,
      "loss": 0.4178,
      "step": 224,
      "task_loss": 0.19036653637886047
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7077669686216033,
      "compression_loss": 0.0,
      "distillation_loss": 0.714052140712738,
      "epoch": 0.21,
      "learning_rate": 4.988817501397505e-05,
      "loss": 0.6902,
      "step": 225,
      "task_loss": 0.4752587080001831
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7078007082986417,
      "compression_loss": 0.0,
      "distillation_loss": 0.36877432465553284,
      "epoch": 0.21,
      "learning_rate": 4.9887166072884e-05,
      "loss": 0.3439,
      "step": 226,
      "task_loss": 0.11986754834651947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7078344397464906,
      "compression_loss": 0.0,
      "distillation_loss": 0.5401185750961304,
      "epoch": 0.22,
      "learning_rate": 4.988615261091938e-05,
      "loss": 0.5173,
      "step": 227,
      "task_loss": 0.31204986572265625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7078681629661537,
      "compression_loss": 0.0,
      "distillation_loss": 0.6624010801315308,
      "epoch": 0.22,
      "learning_rate": 4.9885134628265276e-05,
      "loss": 0.6317,
      "step": 228,
      "task_loss": 0.3551117479801178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7079018779586348,
      "compression_loss": 0.0,
      "distillation_loss": 0.5041351318359375,
      "epoch": 0.22,
      "learning_rate": 4.988411212510663e-05,
      "loss": 0.4863,
      "step": 229,
      "task_loss": 0.3257242739200592
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7079355847249373,
      "compression_loss": 0.0,
      "distillation_loss": 0.5081358551979065,
      "epoch": 0.22,
      "learning_rate": 4.988308510162917e-05,
      "loss": 0.4942,
      "step": 230,
      "task_loss": 0.36878862977027893
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7079692832660651,
      "compression_loss": 0.0,
      "distillation_loss": 0.4084128737449646,
      "epoch": 0.22,
      "learning_rate": 4.988205355801945e-05,
      "loss": 0.3874,
      "step": 231,
      "task_loss": 0.19869351387023926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7080029735830219,
      "compression_loss": 0.0,
      "distillation_loss": 0.6727085113525391,
      "epoch": 0.22,
      "learning_rate": 4.988101749446488e-05,
      "loss": 0.6462,
      "step": 232,
      "task_loss": 0.40740063786506653
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7080366556768113,
      "compression_loss": 0.0,
      "distillation_loss": 0.3574530780315399,
      "epoch": 0.22,
      "learning_rate": 4.987997691115366e-05,
      "loss": 0.3417,
      "step": 233,
      "task_loss": 0.20038984715938568
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.708070329548437,
      "compression_loss": 0.0,
      "distillation_loss": 0.6499674916267395,
      "epoch": 0.22,
      "learning_rate": 4.98789318082748e-05,
      "loss": 0.6204,
      "step": 234,
      "task_loss": 0.3542225658893585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7081039951989025,
      "compression_loss": 0.0,
      "distillation_loss": 0.2822619080543518,
      "epoch": 0.22,
      "learning_rate": 4.987788218601816e-05,
      "loss": 0.2671,
      "step": 235,
      "task_loss": 0.13111086189746857
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7081376526292118,
      "compression_loss": 0.0,
      "distillation_loss": 0.3758002519607544,
      "epoch": 0.22,
      "learning_rate": 4.987682804457441e-05,
      "loss": 0.3537,
      "step": 236,
      "task_loss": 0.15476220846176147
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7081713018403685,
      "compression_loss": 0.0,
      "distillation_loss": 0.42200687527656555,
      "epoch": 0.23,
      "learning_rate": 4.987576938413504e-05,
      "loss": 0.4044,
      "step": 237,
      "task_loss": 0.24637091159820557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7082049428333762,
      "compression_loss": 0.0,
      "distillation_loss": 0.3074836730957031,
      "epoch": 0.23,
      "learning_rate": 4.987470620489235e-05,
      "loss": 0.2978,
      "step": 238,
      "task_loss": 0.21041469275951385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7082385756092385,
      "compression_loss": 0.0,
      "distillation_loss": 0.19442984461784363,
      "epoch": 0.23,
      "learning_rate": 4.9873638507039486e-05,
      "loss": 0.1873,
      "step": 239,
      "task_loss": 0.12338165193796158
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7082722001689592,
      "compression_loss": 0.0,
      "distillation_loss": 0.3584289252758026,
      "epoch": 0.23,
      "learning_rate": 4.987256629077039e-05,
      "loss": 0.34,
      "step": 240,
      "task_loss": 0.17371408641338348
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7083058165135421,
      "compression_loss": 0.0,
      "distillation_loss": 0.8099121451377869,
      "epoch": 0.23,
      "learning_rate": 4.987148955627985e-05,
      "loss": 0.7633,
      "step": 241,
      "task_loss": 0.34426361322402954
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7083394246439906,
      "compression_loss": 0.0,
      "distillation_loss": 0.293813019990921,
      "epoch": 0.23,
      "learning_rate": 4.987040830376344e-05,
      "loss": 0.2808,
      "step": 242,
      "task_loss": 0.16348132491111755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7083730245613087,
      "compression_loss": 0.0,
      "distillation_loss": 0.5357306599617004,
      "epoch": 0.23,
      "learning_rate": 4.9869322533417596e-05,
      "loss": 0.504,
      "step": 243,
      "task_loss": 0.21806874871253967
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7084066162664997,
      "compression_loss": 0.0,
      "distillation_loss": 0.6220539212226868,
      "epoch": 0.23,
      "learning_rate": 4.9868232245439525e-05,
      "loss": 0.5916,
      "step": 244,
      "task_loss": 0.3177708387374878
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7084401997605676,
      "compression_loss": 0.0,
      "distillation_loss": 0.49369296431541443,
      "epoch": 0.23,
      "learning_rate": 4.986713744002731e-05,
      "loss": 0.4649,
      "step": 245,
      "task_loss": 0.20582452416419983
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.708473775044516,
      "compression_loss": 0.0,
      "distillation_loss": 0.46345996856689453,
      "epoch": 0.23,
      "learning_rate": 4.9866038117379824e-05,
      "loss": 0.4486,
      "step": 246,
      "task_loss": 0.31531885266304016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7085073421193486,
      "compression_loss": 0.0,
      "distillation_loss": 0.32199227809906006,
      "epoch": 0.23,
      "learning_rate": 4.986493427769675e-05,
      "loss": 0.3051,
      "step": 247,
      "task_loss": 0.15327921509742737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.708540900986069,
      "compression_loss": 0.0,
      "distillation_loss": 0.4429311156272888,
      "epoch": 0.24,
      "learning_rate": 4.986382592117861e-05,
      "loss": 0.4221,
      "step": 248,
      "task_loss": 0.2345043569803238
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7085744516456809,
      "compression_loss": 0.0,
      "distillation_loss": 0.47061067819595337,
      "epoch": 0.24,
      "learning_rate": 4.986271304802675e-05,
      "loss": 0.4492,
      "step": 249,
      "task_loss": 0.2561434507369995
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7086079940991881,
      "compression_loss": 0.0,
      "distillation_loss": 0.4610680341720581,
      "epoch": 0.24,
      "learning_rate": 4.986159565844333e-05,
      "loss": 0.4471,
      "step": 250,
      "task_loss": 0.32182347774505615
    },
    {
      "epoch": 0.24,
      "eval_accuracy": 0.8532110091743119,
      "eval_loss": 0.6153932809829712,
      "eval_runtime": 14.8163,
      "eval_samples_per_second": 58.854,
      "eval_steps_per_second": 7.357,
      "step": 250
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.708641528347594,
      "compression_loss": 0.0,
      "distillation_loss": 0.53682541847229,
      "epoch": 0.24,
      "learning_rate": 4.986047375263131e-05,
      "loss": 0.513,
      "step": 251,
      "task_loss": 0.2982673645019531
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7086750543919026,
      "compression_loss": 0.0,
      "distillation_loss": 0.29910629987716675,
      "epoch": 0.24,
      "learning_rate": 4.9859347330794515e-05,
      "loss": 0.2856,
      "step": 252,
      "task_loss": 0.16408580541610718
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7087085722331175,
      "compression_loss": 0.0,
      "distillation_loss": 0.49715662002563477,
      "epoch": 0.24,
      "learning_rate": 4.985821639313755e-05,
      "loss": 0.4719,
      "step": 253,
      "task_loss": 0.24493193626403809
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7087420818722422,
      "compression_loss": 0.0,
      "distillation_loss": 0.6056504249572754,
      "epoch": 0.24,
      "learning_rate": 4.985708093986586e-05,
      "loss": 0.5746,
      "step": 254,
      "task_loss": 0.29507917165756226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7087755833102806,
      "compression_loss": 0.0,
      "distillation_loss": 0.5323020219802856,
      "epoch": 0.24,
      "learning_rate": 4.98559409711857e-05,
      "loss": 0.5011,
      "step": 255,
      "task_loss": 0.2197890430688858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7088090765482363,
      "compression_loss": 0.0,
      "distillation_loss": 0.5494606494903564,
      "epoch": 0.24,
      "learning_rate": 4.985479648730416e-05,
      "loss": 0.5286,
      "step": 256,
      "task_loss": 0.3407706320285797
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.708842561587113,
      "compression_loss": 0.0,
      "distillation_loss": 0.35420656204223633,
      "epoch": 0.24,
      "learning_rate": 4.985364748842914e-05,
      "loss": 0.3456,
      "step": 257,
      "task_loss": 0.2684999108314514
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7088760384279144,
      "compression_loss": 0.0,
      "distillation_loss": 0.4922611713409424,
      "epoch": 0.25,
      "learning_rate": 4.985249397476934e-05,
      "loss": 0.4717,
      "step": 258,
      "task_loss": 0.2866893410682678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.708909507071644,
      "compression_loss": 0.0,
      "distillation_loss": 0.34109288454055786,
      "epoch": 0.25,
      "learning_rate": 4.985133594653434e-05,
      "loss": 0.3198,
      "step": 259,
      "task_loss": 0.12784980237483978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7089429675193057,
      "compression_loss": 0.0,
      "distillation_loss": 0.3204500675201416,
      "epoch": 0.25,
      "learning_rate": 4.9850173403934466e-05,
      "loss": 0.3082,
      "step": 260,
      "task_loss": 0.198293536901474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7089764197719032,
      "compression_loss": 0.0,
      "distillation_loss": 0.5372319221496582,
      "epoch": 0.25,
      "learning_rate": 4.9849006347180915e-05,
      "loss": 0.5081,
      "step": 261,
      "task_loss": 0.24620920419692993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.70900986383044,
      "compression_loss": 0.0,
      "distillation_loss": 0.31807684898376465,
      "epoch": 0.25,
      "learning_rate": 4.9847834776485694e-05,
      "loss": 0.3002,
      "step": 262,
      "task_loss": 0.13968217372894287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7090432996959198,
      "compression_loss": 0.0,
      "distillation_loss": 0.3310486078262329,
      "epoch": 0.25,
      "learning_rate": 4.984665869206161e-05,
      "loss": 0.3205,
      "step": 263,
      "task_loss": 0.22510308027267456
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7090767273693465,
      "compression_loss": 0.0,
      "distillation_loss": 0.15543615818023682,
      "epoch": 0.25,
      "learning_rate": 4.984547809412231e-05,
      "loss": 0.15,
      "step": 264,
      "task_loss": 0.10149666666984558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7091101468517236,
      "compression_loss": 0.0,
      "distillation_loss": 0.4099940061569214,
      "epoch": 0.25,
      "learning_rate": 4.984429298288227e-05,
      "loss": 0.385,
      "step": 265,
      "task_loss": 0.1603960394859314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7091435581440548,
      "compression_loss": 0.0,
      "distillation_loss": 0.6284467577934265,
      "epoch": 0.25,
      "learning_rate": 4.984310335855674e-05,
      "loss": 0.5938,
      "step": 266,
      "task_loss": 0.28220003843307495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7091769612473438,
      "compression_loss": 0.0,
      "distillation_loss": 0.4136122167110443,
      "epoch": 0.25,
      "learning_rate": 4.9841909221361855e-05,
      "loss": 0.3988,
      "step": 267,
      "task_loss": 0.26578381657600403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7092103561625943,
      "compression_loss": 0.0,
      "distillation_loss": 0.5570213794708252,
      "epoch": 0.25,
      "learning_rate": 4.9840710571514515e-05,
      "loss": 0.5466,
      "step": 268,
      "task_loss": 0.4525538682937622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7092437428908099,
      "compression_loss": 0.0,
      "distillation_loss": 0.4353582561016083,
      "epoch": 0.26,
      "learning_rate": 4.9839507409232464e-05,
      "loss": 0.4121,
      "step": 269,
      "task_loss": 0.20239636301994324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7092771214329945,
      "compression_loss": 0.0,
      "distillation_loss": 0.5894170999526978,
      "epoch": 0.26,
      "learning_rate": 4.983829973473426e-05,
      "loss": 0.5694,
      "step": 270,
      "task_loss": 0.38912904262542725
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7093104917901515,
      "compression_loss": 0.0,
      "distillation_loss": 0.5201334953308105,
      "epoch": 0.26,
      "learning_rate": 4.983708754823929e-05,
      "loss": 0.4966,
      "step": 271,
      "task_loss": 0.284492552280426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7093438539632848,
      "compression_loss": 0.0,
      "distillation_loss": 0.4675508737564087,
      "epoch": 0.26,
      "learning_rate": 4.983587084996776e-05,
      "loss": 0.4469,
      "step": 272,
      "task_loss": 0.26147744059562683
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.709377207953398,
      "compression_loss": 0.0,
      "distillation_loss": 0.45866286754608154,
      "epoch": 0.26,
      "learning_rate": 4.9834649640140664e-05,
      "loss": 0.4508,
      "step": 273,
      "task_loss": 0.37955737113952637
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7094105537614946,
      "compression_loss": 0.0,
      "distillation_loss": 0.4817972183227539,
      "epoch": 0.26,
      "learning_rate": 4.9833423918979864e-05,
      "loss": 0.4607,
      "step": 274,
      "task_loss": 0.2709713578224182
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7094438913885787,
      "compression_loss": 0.0,
      "distillation_loss": 0.3562793433666229,
      "epoch": 0.26,
      "learning_rate": 4.983219368670801e-05,
      "loss": 0.3413,
      "step": 275,
      "task_loss": 0.20625412464141846
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7094772208356536,
      "compression_loss": 0.0,
      "distillation_loss": 0.45571157336235046,
      "epoch": 0.26,
      "learning_rate": 4.983095894354858e-05,
      "loss": 0.4336,
      "step": 276,
      "task_loss": 0.2345903366804123
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7095105421037231,
      "compression_loss": 0.0,
      "distillation_loss": 0.22800546884536743,
      "epoch": 0.26,
      "learning_rate": 4.9829719689725865e-05,
      "loss": 0.2207,
      "step": 277,
      "task_loss": 0.1548459529876709
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.709543855193791,
      "compression_loss": 0.0,
      "distillation_loss": 0.5845510959625244,
      "epoch": 0.26,
      "learning_rate": 4.982847592546499e-05,
      "loss": 0.564,
      "step": 278,
      "task_loss": 0.3790714740753174
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7095771601068609,
      "compression_loss": 0.0,
      "distillation_loss": 0.2517128884792328,
      "epoch": 0.26,
      "learning_rate": 4.982722765099189e-05,
      "loss": 0.2349,
      "step": 279,
      "task_loss": 0.08311676234006882
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7096104568439363,
      "compression_loss": 0.0,
      "distillation_loss": 0.4391753077507019,
      "epoch": 0.27,
      "learning_rate": 4.982597486653332e-05,
      "loss": 0.4248,
      "step": 280,
      "task_loss": 0.29559123516082764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7096437454060213,
      "compression_loss": 0.0,
      "distillation_loss": 0.5961909294128418,
      "epoch": 0.27,
      "learning_rate": 4.982471757231685e-05,
      "loss": 0.5709,
      "step": 281,
      "task_loss": 0.3435242772102356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7096770257941192,
      "compression_loss": 0.0,
      "distillation_loss": 0.3963918089866638,
      "epoch": 0.27,
      "learning_rate": 4.982345576857087e-05,
      "loss": 0.3802,
      "step": 282,
      "task_loss": 0.23463018238544464
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7097102980092338,
      "compression_loss": 0.0,
      "distillation_loss": 0.40490058064460754,
      "epoch": 0.27,
      "learning_rate": 4.9822189455524604e-05,
      "loss": 0.3843,
      "step": 283,
      "task_loss": 0.1989540457725525
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7097435620523688,
      "compression_loss": 0.0,
      "distillation_loss": 0.40912073850631714,
      "epoch": 0.27,
      "learning_rate": 4.982091863340808e-05,
      "loss": 0.3978,
      "step": 284,
      "task_loss": 0.29543089866638184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7097768179245278,
      "compression_loss": 0.0,
      "distillation_loss": 0.5951185822486877,
      "epoch": 0.27,
      "learning_rate": 4.9819643302452146e-05,
      "loss": 0.5699,
      "step": 285,
      "task_loss": 0.3433946371078491
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7098100656267148,
      "compression_loss": 0.0,
      "distillation_loss": 0.45149531960487366,
      "epoch": 0.27,
      "learning_rate": 4.981836346288847e-05,
      "loss": 0.4407,
      "step": 286,
      "task_loss": 0.3439605236053467
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7098433051599331,
      "compression_loss": 0.0,
      "distillation_loss": 0.9340552687644958,
      "epoch": 0.27,
      "learning_rate": 4.981707911494955e-05,
      "loss": 0.8901,
      "step": 287,
      "task_loss": 0.4947472810745239
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7098765365251865,
      "compression_loss": 0.0,
      "distillation_loss": 0.40463870763778687,
      "epoch": 0.27,
      "learning_rate": 4.981579025886868e-05,
      "loss": 0.379,
      "step": 288,
      "task_loss": 0.14829692244529724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7099097597234787,
      "compression_loss": 0.0,
      "distillation_loss": 0.14002841711044312,
      "epoch": 0.27,
      "learning_rate": 4.981449689488e-05,
      "loss": 0.1338,
      "step": 289,
      "task_loss": 0.07818258553743362
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7099429747558135,
      "compression_loss": 0.0,
      "distillation_loss": 0.45351576805114746,
      "epoch": 0.28,
      "learning_rate": 4.981319902321846e-05,
      "loss": 0.4276,
      "step": 290,
      "task_loss": 0.1947600394487381
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7099761816231944,
      "compression_loss": 0.0,
      "distillation_loss": 0.4904792308807373,
      "epoch": 0.28,
      "learning_rate": 4.981189664411981e-05,
      "loss": 0.4701,
      "step": 291,
      "task_loss": 0.28627243638038635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7100093803266252,
      "compression_loss": 0.0,
      "distillation_loss": 0.6285742521286011,
      "epoch": 0.28,
      "learning_rate": 4.981058975782063e-05,
      "loss": 0.5967,
      "step": 292,
      "task_loss": 0.31022369861602783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7100425708671094,
      "compression_loss": 0.0,
      "distillation_loss": 0.7875237464904785,
      "epoch": 0.28,
      "learning_rate": 4.9809278364558336e-05,
      "loss": 0.7486,
      "step": 293,
      "task_loss": 0.39871400594711304
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.710075753245651,
      "compression_loss": 0.0,
      "distillation_loss": 0.5977748036384583,
      "epoch": 0.28,
      "learning_rate": 4.980796246457115e-05,
      "loss": 0.5728,
      "step": 294,
      "task_loss": 0.3479851484298706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7101089274632534,
      "compression_loss": 0.0,
      "distillation_loss": 0.5126686096191406,
      "epoch": 0.28,
      "learning_rate": 4.9806642058098105e-05,
      "loss": 0.4815,
      "step": 295,
      "task_loss": 0.20053212344646454
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7101420935209204,
      "compression_loss": 0.0,
      "distillation_loss": 0.5196264982223511,
      "epoch": 0.28,
      "learning_rate": 4.980531714537905e-05,
      "loss": 0.4878,
      "step": 296,
      "task_loss": 0.2012956738471985
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7101752514196557,
      "compression_loss": 0.0,
      "distillation_loss": 0.33446091413497925,
      "epoch": 0.28,
      "learning_rate": 4.980398772665468e-05,
      "loss": 0.3161,
      "step": 297,
      "task_loss": 0.15055128931999207
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7102084011604629,
      "compression_loss": 0.0,
      "distillation_loss": 0.6768823862075806,
      "epoch": 0.28,
      "learning_rate": 4.980265380216649e-05,
      "loss": 0.6398,
      "step": 298,
      "task_loss": 0.30652397871017456
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7102415427443457,
      "compression_loss": 0.0,
      "distillation_loss": 0.3801252245903015,
      "epoch": 0.28,
      "learning_rate": 4.9801315372156775e-05,
      "loss": 0.3637,
      "step": 299,
      "task_loss": 0.21593311429023743
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7102746761723079,
      "compression_loss": 0.0,
      "distillation_loss": 0.3485510051250458,
      "epoch": 0.28,
      "learning_rate": 4.979997243686868e-05,
      "loss": 0.3279,
      "step": 300,
      "task_loss": 0.14158813655376434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7103078014453531,
      "compression_loss": 0.0,
      "distillation_loss": 0.4838918447494507,
      "epoch": 0.29,
      "learning_rate": 4.979862499654615e-05,
      "loss": 0.4596,
      "step": 301,
      "task_loss": 0.24049346148967743
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7103409185644849,
      "compression_loss": 0.0,
      "distillation_loss": 0.5298963785171509,
      "epoch": 0.29,
      "learning_rate": 4.9797273051433966e-05,
      "loss": 0.4982,
      "step": 302,
      "task_loss": 0.21265023946762085
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7103740275307071,
      "compression_loss": 0.0,
      "distillation_loss": 0.1947401762008667,
      "epoch": 0.29,
      "learning_rate": 4.97959166017777e-05,
      "loss": 0.1844,
      "step": 303,
      "task_loss": 0.0909617617726326
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7104071283450233,
      "compression_loss": 0.0,
      "distillation_loss": 0.3400229215621948,
      "epoch": 0.29,
      "learning_rate": 4.979455564782377e-05,
      "loss": 0.3316,
      "step": 304,
      "task_loss": 0.2561742961406708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7104402210084373,
      "compression_loss": 0.0,
      "distillation_loss": 0.33108600974082947,
      "epoch": 0.29,
      "learning_rate": 4.9793190189819395e-05,
      "loss": 0.3223,
      "step": 305,
      "task_loss": 0.24314963817596436
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7104733055219526,
      "compression_loss": 0.0,
      "distillation_loss": 0.270094633102417,
      "epoch": 0.29,
      "learning_rate": 4.979182022801262e-05,
      "loss": 0.2599,
      "step": 306,
      "task_loss": 0.1684069037437439
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7105063818865731,
      "compression_loss": 0.0,
      "distillation_loss": 0.3968997001647949,
      "epoch": 0.29,
      "learning_rate": 4.979044576265229e-05,
      "loss": 0.3775,
      "step": 307,
      "task_loss": 0.203133687376976
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7105394501033022,
      "compression_loss": 0.0,
      "distillation_loss": 0.4618909955024719,
      "epoch": 0.29,
      "learning_rate": 4.9789066793988106e-05,
      "loss": 0.4387,
      "step": 308,
      "task_loss": 0.23047326505184174
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7105725101731439,
      "compression_loss": 0.0,
      "distillation_loss": 0.4338553547859192,
      "epoch": 0.29,
      "learning_rate": 4.978768332227054e-05,
      "loss": 0.4193,
      "step": 309,
      "task_loss": 0.2878293991088867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7106055620971018,
      "compression_loss": 0.0,
      "distillation_loss": 0.6526356935501099,
      "epoch": 0.29,
      "learning_rate": 4.9786295347750936e-05,
      "loss": 0.6164,
      "step": 310,
      "task_loss": 0.29025229811668396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7106386058761793,
      "compression_loss": 0.0,
      "distillation_loss": 0.7481920123100281,
      "epoch": 0.3,
      "learning_rate": 4.9784902870681406e-05,
      "loss": 0.7157,
      "step": 311,
      "task_loss": 0.4227951765060425
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7106716415113805,
      "compression_loss": 0.0,
      "distillation_loss": 0.679728627204895,
      "epoch": 0.3,
      "learning_rate": 4.97835058913149e-05,
      "loss": 0.6661,
      "step": 312,
      "task_loss": 0.5437166690826416
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7107046690037088,
      "compression_loss": 0.0,
      "distillation_loss": 0.6855102777481079,
      "epoch": 0.3,
      "learning_rate": 4.9782104409905186e-05,
      "loss": 0.6504,
      "step": 313,
      "task_loss": 0.33481907844543457
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.710737688354168,
      "compression_loss": 0.0,
      "distillation_loss": 0.4069738984107971,
      "epoch": 0.3,
      "learning_rate": 4.9780698426706864e-05,
      "loss": 0.3808,
      "step": 314,
      "task_loss": 0.14478835463523865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7107706995637616,
      "compression_loss": 0.0,
      "distillation_loss": 0.6615716218948364,
      "epoch": 0.3,
      "learning_rate": 4.977928794197532e-05,
      "loss": 0.6309,
      "step": 315,
      "task_loss": 0.3549352288246155
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7108037026334935,
      "compression_loss": 0.0,
      "distillation_loss": 0.5655452609062195,
      "epoch": 0.3,
      "learning_rate": 4.9777872955966785e-05,
      "loss": 0.5316,
      "step": 316,
      "task_loss": 0.22563889622688293
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7108366975643673,
      "compression_loss": 0.0,
      "distillation_loss": 0.3800777792930603,
      "epoch": 0.3,
      "learning_rate": 4.97764534689383e-05,
      "loss": 0.3621,
      "step": 317,
      "task_loss": 0.20060645043849945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7108696843573867,
      "compression_loss": 0.0,
      "distillation_loss": 0.2623631954193115,
      "epoch": 0.3,
      "learning_rate": 4.977502948114772e-05,
      "loss": 0.2563,
      "step": 318,
      "task_loss": 0.20183011889457703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7109026630135554,
      "compression_loss": 0.0,
      "distillation_loss": 0.4012034833431244,
      "epoch": 0.3,
      "learning_rate": 4.977360099285371e-05,
      "loss": 0.3783,
      "step": 319,
      "task_loss": 0.17202433943748474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7109356335338771,
      "compression_loss": 0.0,
      "distillation_loss": 0.35331350564956665,
      "epoch": 0.3,
      "learning_rate": 4.9772168004315765e-05,
      "loss": 0.3308,
      "step": 320,
      "task_loss": 0.12813310325145721
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7109685959193554,
      "compression_loss": 0.0,
      "distillation_loss": 0.3918440043926239,
      "epoch": 0.3,
      "learning_rate": 4.9770730515794204e-05,
      "loss": 0.3696,
      "step": 321,
      "task_loss": 0.16927433013916016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.711001550170994,
      "compression_loss": 0.0,
      "distillation_loss": 0.3784354627132416,
      "epoch": 0.31,
      "learning_rate": 4.976928852755015e-05,
      "loss": 0.3608,
      "step": 322,
      "task_loss": 0.2020285427570343
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7110344962897965,
      "compression_loss": 0.0,
      "distillation_loss": 0.5241067409515381,
      "epoch": 0.31,
      "learning_rate": 4.976784203984554e-05,
      "loss": 0.501,
      "step": 323,
      "task_loss": 0.2931115925312042
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7110674342767669,
      "compression_loss": 0.0,
      "distillation_loss": 0.3190663456916809,
      "epoch": 0.31,
      "learning_rate": 4.976639105294314e-05,
      "loss": 0.308,
      "step": 324,
      "task_loss": 0.2081044614315033
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7111003641329084,
      "compression_loss": 0.0,
      "distillation_loss": 0.3578122854232788,
      "epoch": 0.31,
      "learning_rate": 4.976493556710653e-05,
      "loss": 0.3369,
      "step": 325,
      "task_loss": 0.14846715331077576
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7111332858592252,
      "compression_loss": 0.0,
      "distillation_loss": 0.4352418780326843,
      "epoch": 0.31,
      "learning_rate": 4.976347558260011e-05,
      "loss": 0.4134,
      "step": 326,
      "task_loss": 0.21667122840881348
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7111661994567205,
      "compression_loss": 0.0,
      "distillation_loss": 0.36462095379829407,
      "epoch": 0.31,
      "learning_rate": 4.976201109968908e-05,
      "loss": 0.3506,
      "step": 327,
      "task_loss": 0.2242008000612259
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7111991049263984,
      "compression_loss": 0.0,
      "distillation_loss": 0.5992603302001953,
      "epoch": 0.31,
      "learning_rate": 4.976054211863949e-05,
      "loss": 0.5782,
      "step": 328,
      "task_loss": 0.38846805691719055
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7112320022692623,
      "compression_loss": 0.0,
      "distillation_loss": 0.1498991847038269,
      "epoch": 0.31,
      "learning_rate": 4.9759068639718166e-05,
      "loss": 0.1485,
      "step": 329,
      "task_loss": 0.13616728782653809
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.711264891486316,
      "compression_loss": 0.0,
      "distillation_loss": 0.7137056589126587,
      "epoch": 0.31,
      "learning_rate": 4.975759066319278e-05,
      "loss": 0.6852,
      "step": 330,
      "task_loss": 0.4287063479423523
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7112977725785632,
      "compression_loss": 0.0,
      "distillation_loss": 0.3997446298599243,
      "epoch": 0.31,
      "learning_rate": 4.9756108189331825e-05,
      "loss": 0.385,
      "step": 331,
      "task_loss": 0.2526022791862488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7113306455470074,
      "compression_loss": 0.0,
      "distillation_loss": 0.6099421977996826,
      "epoch": 0.32,
      "learning_rate": 4.975462121840458e-05,
      "loss": 0.5846,
      "step": 332,
      "task_loss": 0.35605889558792114
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7113635103926526,
      "compression_loss": 0.0,
      "distillation_loss": 0.3887147605419159,
      "epoch": 0.32,
      "learning_rate": 4.975312975068118e-05,
      "loss": 0.3771,
      "step": 333,
      "task_loss": 0.27239084243774414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7113963671165022,
      "compression_loss": 0.0,
      "distillation_loss": 0.3457412123680115,
      "epoch": 0.32,
      "learning_rate": 4.975163378643255e-05,
      "loss": 0.3356,
      "step": 334,
      "task_loss": 0.2438831776380539
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7114292157195601,
      "compression_loss": 0.0,
      "distillation_loss": 0.26972734928131104,
      "epoch": 0.32,
      "learning_rate": 4.975013332593044e-05,
      "loss": 0.2563,
      "step": 335,
      "task_loss": 0.13514116406440735
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7114620562028297,
      "compression_loss": 0.0,
      "distillation_loss": 0.269406259059906,
      "epoch": 0.32,
      "learning_rate": 4.97486283694474e-05,
      "loss": 0.2526,
      "step": 336,
      "task_loss": 0.10177846252918243
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7114948885673149,
      "compression_loss": 0.0,
      "distillation_loss": 0.5931138396263123,
      "epoch": 0.32,
      "learning_rate": 4.974711891725684e-05,
      "loss": 0.5679,
      "step": 337,
      "task_loss": 0.34124326705932617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7115277128140194,
      "compression_loss": 0.0,
      "distillation_loss": 0.12281601130962372,
      "epoch": 0.32,
      "learning_rate": 4.9745604969632934e-05,
      "loss": 0.1292,
      "step": 338,
      "task_loss": 0.18707461655139923
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7115605289439467,
      "compression_loss": 0.0,
      "distillation_loss": 0.6009297966957092,
      "epoch": 0.32,
      "learning_rate": 4.974408652685072e-05,
      "loss": 0.5704,
      "step": 339,
      "task_loss": 0.296112984418869
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7115933369581007,
      "compression_loss": 0.0,
      "distillation_loss": 0.6147419214248657,
      "epoch": 0.32,
      "learning_rate": 4.974256358918601e-05,
      "loss": 0.586,
      "step": 340,
      "task_loss": 0.3271123766899109
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7116261368574849,
      "compression_loss": 0.0,
      "distillation_loss": 0.5920723080635071,
      "epoch": 0.32,
      "learning_rate": 4.9741036156915464e-05,
      "loss": 0.565,
      "step": 341,
      "task_loss": 0.32182377576828003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7116589286431031,
      "compression_loss": 0.0,
      "distillation_loss": 0.2902904450893402,
      "epoch": 0.32,
      "learning_rate": 4.973950423031655e-05,
      "loss": 0.2809,
      "step": 342,
      "task_loss": 0.19625771045684814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.711691712315959,
      "compression_loss": 0.0,
      "distillation_loss": 0.674541711807251,
      "epoch": 0.33,
      "learning_rate": 4.9737967809667546e-05,
      "loss": 0.6477,
      "step": 343,
      "task_loss": 0.4057275056838989
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7117244878770561,
      "compression_loss": 0.0,
      "distillation_loss": 0.18200403451919556,
      "epoch": 0.33,
      "learning_rate": 4.9736426895247545e-05,
      "loss": 0.1747,
      "step": 344,
      "task_loss": 0.10881594568490982
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7117572553273983,
      "compression_loss": 0.0,
      "distillation_loss": 0.5339623093605042,
      "epoch": 0.33,
      "learning_rate": 4.973488148733647e-05,
      "loss": 0.5104,
      "step": 345,
      "task_loss": 0.29824596643447876
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7117900146679892,
      "compression_loss": 0.0,
      "distillation_loss": 0.5717617273330688,
      "epoch": 0.33,
      "learning_rate": 4.973333158621505e-05,
      "loss": 0.553,
      "step": 346,
      "task_loss": 0.3842325806617737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7118227658998324,
      "compression_loss": 0.0,
      "distillation_loss": 0.4612230062484741,
      "epoch": 0.33,
      "learning_rate": 4.973177719216483e-05,
      "loss": 0.4341,
      "step": 347,
      "task_loss": 0.1896570473909378
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7118555090239316,
      "compression_loss": 0.0,
      "distillation_loss": 0.383368581533432,
      "epoch": 0.33,
      "learning_rate": 4.973021830546817e-05,
      "loss": 0.3685,
      "step": 348,
      "task_loss": 0.23502130806446075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7118882440412907,
      "compression_loss": 0.0,
      "distillation_loss": 0.33486291766166687,
      "epoch": 0.33,
      "learning_rate": 4.972865492640826e-05,
      "loss": 0.3254,
      "step": 349,
      "task_loss": 0.24004624783992767
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7119209709529131,
      "compression_loss": 0.0,
      "distillation_loss": 0.37039846181869507,
      "epoch": 0.33,
      "learning_rate": 4.972708705526908e-05,
      "loss": 0.3512,
      "step": 350,
      "task_loss": 0.1785244345664978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7119536897598027,
      "compression_loss": 0.0,
      "distillation_loss": 0.4161568284034729,
      "epoch": 0.33,
      "learning_rate": 4.972551469233545e-05,
      "loss": 0.3904,
      "step": 351,
      "task_loss": 0.15905624628067017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7119864004629629,
      "compression_loss": 0.0,
      "distillation_loss": 0.4145691394805908,
      "epoch": 0.33,
      "learning_rate": 4.9723937837892996e-05,
      "loss": 0.3912,
      "step": 352,
      "task_loss": 0.18065348267555237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7120191030633978,
      "compression_loss": 0.0,
      "distillation_loss": 0.262037992477417,
      "epoch": 0.34,
      "learning_rate": 4.972235649222817e-05,
      "loss": 0.2493,
      "step": 353,
      "task_loss": 0.13479451835155487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7120517975621107,
      "compression_loss": 0.0,
      "distillation_loss": 0.3872963786125183,
      "epoch": 0.34,
      "learning_rate": 4.972077065562821e-05,
      "loss": 0.3693,
      "step": 354,
      "task_loss": 0.2070463001728058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7120844839601054,
      "compression_loss": 0.0,
      "distillation_loss": 0.48908573389053345,
      "epoch": 0.34,
      "learning_rate": 4.971918032838122e-05,
      "loss": 0.4654,
      "step": 355,
      "task_loss": 0.2519568204879761
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7121171622583856,
      "compression_loss": 0.0,
      "distillation_loss": 0.15656299889087677,
      "epoch": 0.34,
      "learning_rate": 4.9717585510776065e-05,
      "loss": 0.1556,
      "step": 356,
      "task_loss": 0.1469275802373886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.712149832457955,
      "compression_loss": 0.0,
      "distillation_loss": 0.42582622170448303,
      "epoch": 0.34,
      "learning_rate": 4.971598620310246e-05,
      "loss": 0.4109,
      "step": 357,
      "task_loss": 0.27654868364334106
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7121824945598174,
      "compression_loss": 0.0,
      "distillation_loss": 0.4828101396560669,
      "epoch": 0.34,
      "learning_rate": 4.9714382405650926e-05,
      "loss": 0.4592,
      "step": 358,
      "task_loss": 0.24672147631645203
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7122151485649763,
      "compression_loss": 0.0,
      "distillation_loss": 0.2761083245277405,
      "epoch": 0.34,
      "learning_rate": 4.971277411871281e-05,
      "loss": 0.266,
      "step": 359,
      "task_loss": 0.17460951209068298
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7122477944744354,
      "compression_loss": 0.0,
      "distillation_loss": 0.5871754884719849,
      "epoch": 0.34,
      "learning_rate": 4.971116134258025e-05,
      "loss": 0.5566,
      "step": 360,
      "task_loss": 0.28161656856536865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7122804322891985,
      "compression_loss": 0.0,
      "distillation_loss": 0.4658134877681732,
      "epoch": 0.34,
      "learning_rate": 4.9709544077546235e-05,
      "loss": 0.4412,
      "step": 361,
      "task_loss": 0.21951305866241455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7123130620102691,
      "compression_loss": 0.0,
      "distillation_loss": 0.4061258137226105,
      "epoch": 0.34,
      "learning_rate": 4.9707922323904524e-05,
      "loss": 0.3812,
      "step": 362,
      "task_loss": 0.1569288671016693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.712345683638651,
      "compression_loss": 0.0,
      "distillation_loss": 0.4409645199775696,
      "epoch": 0.34,
      "learning_rate": 4.9706296081949724e-05,
      "loss": 0.4208,
      "step": 363,
      "task_loss": 0.23960046470165253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7123782971753478,
      "compression_loss": 0.0,
      "distillation_loss": 0.13879182934761047,
      "epoch": 0.35,
      "learning_rate": 4.9704665351977266e-05,
      "loss": 0.1371,
      "step": 364,
      "task_loss": 0.12139101326465607
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7124109026213634,
      "compression_loss": 0.0,
      "distillation_loss": 0.5069670677185059,
      "epoch": 0.35,
      "learning_rate": 4.9703030134283356e-05,
      "loss": 0.4817,
      "step": 365,
      "task_loss": 0.25386273860931396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7124434999777013,
      "compression_loss": 0.0,
      "distillation_loss": 0.46004921197891235,
      "epoch": 0.35,
      "learning_rate": 4.970139042916506e-05,
      "loss": 0.4276,
      "step": 366,
      "task_loss": 0.13600590825080872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7124760892453651,
      "compression_loss": 0.0,
      "distillation_loss": 0.47816845774650574,
      "epoch": 0.35,
      "learning_rate": 4.969974623692023e-05,
      "loss": 0.4559,
      "step": 367,
      "task_loss": 0.2554030418395996
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7125086704253587,
      "compression_loss": 0.0,
      "distillation_loss": 0.5051754117012024,
      "epoch": 0.35,
      "learning_rate": 4.969809755784753e-05,
      "loss": 0.4801,
      "step": 368,
      "task_loss": 0.2545245885848999
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7125412435186856,
      "compression_loss": 0.0,
      "distillation_loss": 0.5262600183486938,
      "epoch": 0.35,
      "learning_rate": 4.969644439224647e-05,
      "loss": 0.5022,
      "step": 369,
      "task_loss": 0.28538602590560913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7125738085263497,
      "compression_loss": 0.0,
      "distillation_loss": 0.32377320528030396,
      "epoch": 0.35,
      "learning_rate": 4.969478674041735e-05,
      "loss": 0.3042,
      "step": 370,
      "task_loss": 0.12757205963134766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7126063654493544,
      "compression_loss": 0.0,
      "distillation_loss": 0.31874191761016846,
      "epoch": 0.35,
      "learning_rate": 4.969312460266128e-05,
      "loss": 0.3064,
      "step": 371,
      "task_loss": 0.19505202770233154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7126389142887035,
      "compression_loss": 0.0,
      "distillation_loss": 0.3652954697608948,
      "epoch": 0.35,
      "learning_rate": 4.969145797928021e-05,
      "loss": 0.3444,
      "step": 372,
      "task_loss": 0.156343013048172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7126714550454009,
      "compression_loss": 0.0,
      "distillation_loss": 0.19748246669769287,
      "epoch": 0.35,
      "learning_rate": 4.968978687057687e-05,
      "loss": 0.1868,
      "step": 373,
      "task_loss": 0.09101182222366333
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7127039877204498,
      "compression_loss": 0.0,
      "distillation_loss": 0.5082756876945496,
      "epoch": 0.36,
      "learning_rate": 4.9688111276854846e-05,
      "loss": 0.4828,
      "step": 374,
      "task_loss": 0.2530589699745178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7127365123148545,
      "compression_loss": 0.0,
      "distillation_loss": 0.48118260502815247,
      "epoch": 0.36,
      "learning_rate": 4.9686431198418515e-05,
      "loss": 0.4566,
      "step": 375,
      "task_loss": 0.23539604246616364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7127690288296181,
      "compression_loss": 0.0,
      "distillation_loss": 0.43502384424209595,
      "epoch": 0.36,
      "learning_rate": 4.968474663557306e-05,
      "loss": 0.4128,
      "step": 376,
      "task_loss": 0.21273593604564667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7128015372657446,
      "compression_loss": 0.0,
      "distillation_loss": 0.4908156096935272,
      "epoch": 0.36,
      "learning_rate": 4.9683057588624494e-05,
      "loss": 0.4682,
      "step": 377,
      "task_loss": 0.26497960090637207
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7128340376242377,
      "compression_loss": 0.0,
      "distillation_loss": 0.5738304853439331,
      "epoch": 0.36,
      "learning_rate": 4.968136405787964e-05,
      "loss": 0.5737,
      "step": 378,
      "task_loss": 0.5724409818649292
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7128665299061009,
      "compression_loss": 0.0,
      "distillation_loss": 0.31743335723876953,
      "epoch": 0.36,
      "learning_rate": 4.967966604364614e-05,
      "loss": 0.3031,
      "step": 379,
      "task_loss": 0.17429381608963013
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.712899014112338,
      "compression_loss": 0.0,
      "distillation_loss": 0.6811038851737976,
      "epoch": 0.36,
      "learning_rate": 4.9677963546232445e-05,
      "loss": 0.651,
      "step": 380,
      "task_loss": 0.3800850212574005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7129314902439526,
      "compression_loss": 0.0,
      "distillation_loss": 0.5749964714050293,
      "epoch": 0.36,
      "learning_rate": 4.967625656594782e-05,
      "loss": 0.5545,
      "step": 381,
      "task_loss": 0.3700599670410156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7129639583019486,
      "compression_loss": 0.0,
      "distillation_loss": 0.5684269666671753,
      "epoch": 0.36,
      "learning_rate": 4.967454510310235e-05,
      "loss": 0.5486,
      "step": 382,
      "task_loss": 0.3700346052646637
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7129964182873294,
      "compression_loss": 0.0,
      "distillation_loss": 0.42811310291290283,
      "epoch": 0.36,
      "learning_rate": 4.967282915800693e-05,
      "loss": 0.4135,
      "step": 383,
      "task_loss": 0.2819935083389282
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7130288702010988,
      "compression_loss": 0.0,
      "distillation_loss": 0.5204276442527771,
      "epoch": 0.36,
      "learning_rate": 4.9671108730973274e-05,
      "loss": 0.4932,
      "step": 384,
      "task_loss": 0.24831140041351318
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7130613140442605,
      "compression_loss": 0.0,
      "distillation_loss": 0.33399271965026855,
      "epoch": 0.37,
      "learning_rate": 4.9669383822313886e-05,
      "loss": 0.3222,
      "step": 385,
      "task_loss": 0.21643346548080444
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7130937498178183,
      "compression_loss": 0.0,
      "distillation_loss": 0.318186491727829,
      "epoch": 0.37,
      "learning_rate": 4.966765443234212e-05,
      "loss": 0.3105,
      "step": 386,
      "task_loss": 0.24149882793426514
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7131261775227756,
      "compression_loss": 0.0,
      "distillation_loss": 0.20772957801818848,
      "epoch": 0.37,
      "learning_rate": 4.966592056137213e-05,
      "loss": 0.2072,
      "step": 387,
      "task_loss": 0.20199303328990936
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7131585971601364,
      "compression_loss": 0.0,
      "distillation_loss": 0.29850098490715027,
      "epoch": 0.37,
      "learning_rate": 4.966418220971888e-05,
      "loss": 0.2825,
      "step": 388,
      "task_loss": 0.13806317746639252
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.713191008730904,
      "compression_loss": 0.0,
      "distillation_loss": 0.555107831954956,
      "epoch": 0.37,
      "learning_rate": 4.9662439377698145e-05,
      "loss": 0.5276,
      "step": 389,
      "task_loss": 0.2797929644584656
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7132234122360824,
      "compression_loss": 0.0,
      "distillation_loss": 0.31871169805526733,
      "epoch": 0.37,
      "learning_rate": 4.9660692065626515e-05,
      "loss": 0.3048,
      "step": 390,
      "task_loss": 0.17996791005134583
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7132558076766753,
      "compression_loss": 0.0,
      "distillation_loss": 0.34103503823280334,
      "epoch": 0.37,
      "learning_rate": 4.965894027382141e-05,
      "loss": 0.3362,
      "step": 391,
      "task_loss": 0.29244738817214966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7132881950536861,
      "compression_loss": 0.0,
      "distillation_loss": 0.42130306363105774,
      "epoch": 0.37,
      "learning_rate": 4.965718400260105e-05,
      "loss": 0.4119,
      "step": 392,
      "task_loss": 0.32684704661369324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7133205743681188,
      "compression_loss": 0.0,
      "distillation_loss": 0.252642959356308,
      "epoch": 0.37,
      "learning_rate": 4.965542325228446e-05,
      "loss": 0.2442,
      "step": 393,
      "task_loss": 0.16772150993347168
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7133529456209768,
      "compression_loss": 0.0,
      "distillation_loss": 0.18364742398262024,
      "epoch": 0.37,
      "learning_rate": 4.96536580231915e-05,
      "loss": 0.172,
      "step": 394,
      "task_loss": 0.06767023354768753
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7133853088132639,
      "compression_loss": 0.0,
      "distillation_loss": 0.40755707025527954,
      "epoch": 0.38,
      "learning_rate": 4.9651888315642815e-05,
      "loss": 0.3952,
      "step": 395,
      "task_loss": 0.2839694619178772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7134176639459838,
      "compression_loss": 0.0,
      "distillation_loss": 0.5597716569900513,
      "epoch": 0.38,
      "learning_rate": 4.96501141299599e-05,
      "loss": 0.5393,
      "step": 396,
      "task_loss": 0.35467255115509033
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7134500110201403,
      "compression_loss": 0.0,
      "distillation_loss": 0.44882553815841675,
      "epoch": 0.38,
      "learning_rate": 4.9648335466465035e-05,
      "loss": 0.4253,
      "step": 397,
      "task_loss": 0.21373212337493896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7134823500367369,
      "compression_loss": 0.0,
      "distillation_loss": 0.562800407409668,
      "epoch": 0.38,
      "learning_rate": 4.964655232548133e-05,
      "loss": 0.5389,
      "step": 398,
      "task_loss": 0.3235562741756439
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7135146809967773,
      "compression_loss": 0.0,
      "distillation_loss": 0.43563634157180786,
      "epoch": 0.38,
      "learning_rate": 4.964476470733269e-05,
      "loss": 0.4108,
      "step": 399,
      "task_loss": 0.18746916949748993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7135470039012652,
      "compression_loss": 0.0,
      "distillation_loss": 0.5361025333404541,
      "epoch": 0.38,
      "learning_rate": 4.964297261234385e-05,
      "loss": 0.5201,
      "step": 400,
      "task_loss": 0.3763747215270996
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7135793187512043,
      "compression_loss": 0.0,
      "distillation_loss": 0.40355557203292847,
      "epoch": 0.38,
      "learning_rate": 4.964117604084036e-05,
      "loss": 0.3828,
      "step": 401,
      "task_loss": 0.19647303223609924
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7136116255475984,
      "compression_loss": 0.0,
      "distillation_loss": 0.4803870916366577,
      "epoch": 0.38,
      "learning_rate": 4.963937499314857e-05,
      "loss": 0.4581,
      "step": 402,
      "task_loss": 0.25726521015167236
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.713643924291451,
      "compression_loss": 0.0,
      "distillation_loss": 0.6330090761184692,
      "epoch": 0.38,
      "learning_rate": 4.963756946959564e-05,
      "loss": 0.6021,
      "step": 403,
      "task_loss": 0.3235839009284973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7136762149837658,
      "compression_loss": 0.0,
      "distillation_loss": 0.40180158615112305,
      "epoch": 0.38,
      "learning_rate": 4.9635759470509554e-05,
      "loss": 0.3963,
      "step": 404,
      "task_loss": 0.3472459614276886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7137084976255466,
      "compression_loss": 0.0,
      "distillation_loss": 0.29341554641723633,
      "epoch": 0.38,
      "learning_rate": 4.9633944996219125e-05,
      "loss": 0.2771,
      "step": 405,
      "task_loss": 0.12987171113491058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.713740772217797,
      "compression_loss": 0.0,
      "distillation_loss": 0.23439250886440277,
      "epoch": 0.39,
      "learning_rate": 4.9632126047053954e-05,
      "loss": 0.2184,
      "step": 406,
      "task_loss": 0.07485216110944748
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7137730387615207,
      "compression_loss": 0.0,
      "distillation_loss": 0.2850918769836426,
      "epoch": 0.39,
      "learning_rate": 4.963030262334445e-05,
      "loss": 0.2713,
      "step": 407,
      "task_loss": 0.14699222147464752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7138052972577213,
      "compression_loss": 0.0,
      "distillation_loss": 0.392857164144516,
      "epoch": 0.39,
      "learning_rate": 4.962847472542185e-05,
      "loss": 0.3697,
      "step": 408,
      "task_loss": 0.16155928373336792
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7138375477074026,
      "compression_loss": 0.0,
      "distillation_loss": 0.3292420208454132,
      "epoch": 0.39,
      "learning_rate": 4.96266423536182e-05,
      "loss": 0.3101,
      "step": 409,
      "task_loss": 0.13783293962478638
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7138697901115683,
      "compression_loss": 0.0,
      "distillation_loss": 0.5401057600975037,
      "epoch": 0.39,
      "learning_rate": 4.9624805508266375e-05,
      "loss": 0.5129,
      "step": 410,
      "task_loss": 0.26759254932403564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.713902024471222,
      "compression_loss": 0.0,
      "distillation_loss": 0.4787713289260864,
      "epoch": 0.39,
      "learning_rate": 4.9622964189700026e-05,
      "loss": 0.4574,
      "step": 411,
      "task_loss": 0.2650720477104187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7139342507873674,
      "compression_loss": 0.0,
      "distillation_loss": 0.3292551040649414,
      "epoch": 0.39,
      "learning_rate": 4.962111839825365e-05,
      "loss": 0.3079,
      "step": 412,
      "task_loss": 0.11532945930957794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7139664690610081,
      "compression_loss": 0.0,
      "distillation_loss": 0.32782095670700073,
      "epoch": 0.39,
      "learning_rate": 4.961926813426254e-05,
      "loss": 0.3213,
      "step": 413,
      "task_loss": 0.2627405524253845
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.713998679293148,
      "compression_loss": 0.0,
      "distillation_loss": 0.4790850579738617,
      "epoch": 0.39,
      "learning_rate": 4.9617413398062814e-05,
      "loss": 0.4547,
      "step": 414,
      "task_loss": 0.23543286323547363
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7140308814847907,
      "compression_loss": 0.0,
      "distillation_loss": 0.2713318467140198,
      "epoch": 0.39,
      "learning_rate": 4.9615554189991374e-05,
      "loss": 0.2603,
      "step": 415,
      "task_loss": 0.16122034192085266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7140630756369397,
      "compression_loss": 0.0,
      "distillation_loss": 0.32149723172187805,
      "epoch": 0.4,
      "learning_rate": 4.9613690510385965e-05,
      "loss": 0.303,
      "step": 416,
      "task_loss": 0.13658498227596283
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7140952617505988,
      "compression_loss": 0.0,
      "distillation_loss": 0.22573064267635345,
      "epoch": 0.4,
      "learning_rate": 4.961182235958515e-05,
      "loss": 0.2124,
      "step": 417,
      "task_loss": 0.09292272478342056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7141274398267717,
      "compression_loss": 0.0,
      "distillation_loss": 0.7057199478149414,
      "epoch": 0.4,
      "learning_rate": 4.9609949737928254e-05,
      "loss": 0.6774,
      "step": 418,
      "task_loss": 0.42263519763946533
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7141596098664622,
      "compression_loss": 0.0,
      "distillation_loss": 0.2580406963825226,
      "epoch": 0.4,
      "learning_rate": 4.9608072645755476e-05,
      "loss": 0.2494,
      "step": 419,
      "task_loss": 0.17207808792591095
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7141917718706738,
      "compression_loss": 0.0,
      "distillation_loss": 0.297884464263916,
      "epoch": 0.4,
      "learning_rate": 4.960619108340778e-05,
      "loss": 0.2873,
      "step": 420,
      "task_loss": 0.19168856739997864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7142239258404103,
      "compression_loss": 0.0,
      "distillation_loss": 0.10459847003221512,
      "epoch": 0.4,
      "learning_rate": 4.9604305051226976e-05,
      "loss": 0.0963,
      "step": 421,
      "task_loss": 0.0215504951775074
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7142560717766753,
      "compression_loss": 0.0,
      "distillation_loss": 0.2625170648097992,
      "epoch": 0.4,
      "learning_rate": 4.960241454955566e-05,
      "loss": 0.2528,
      "step": 422,
      "task_loss": 0.1657557189464569
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7142882096804724,
      "compression_loss": 0.0,
      "distillation_loss": 0.49204713106155396,
      "epoch": 0.4,
      "learning_rate": 4.960051957873725e-05,
      "loss": 0.4776,
      "step": 423,
      "task_loss": 0.34725382924079895
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7143203395528055,
      "compression_loss": 0.0,
      "distillation_loss": 0.20165809988975525,
      "epoch": 0.4,
      "learning_rate": 4.959862013911599e-05,
      "loss": 0.1876,
      "step": 424,
      "task_loss": 0.06141514703631401
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7143524613946781,
      "compression_loss": 0.0,
      "distillation_loss": 0.24638772010803223,
      "epoch": 0.4,
      "learning_rate": 4.959671623103691e-05,
      "loss": 0.2331,
      "step": 425,
      "task_loss": 0.113340824842453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7143845752070941,
      "compression_loss": 0.0,
      "distillation_loss": 0.7320840358734131,
      "epoch": 0.4,
      "learning_rate": 4.959480785484587e-05,
      "loss": 0.7126,
      "step": 426,
      "task_loss": 0.53769451379776
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7144166809910568,
      "compression_loss": 0.0,
      "distillation_loss": 0.3415003716945648,
      "epoch": 0.41,
      "learning_rate": 4.959289501088953e-05,
      "loss": 0.3285,
      "step": 427,
      "task_loss": 0.21169662475585938
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7144487787475703,
      "compression_loss": 0.0,
      "distillation_loss": 0.3320407271385193,
      "epoch": 0.41,
      "learning_rate": 4.9590977699515374e-05,
      "loss": 0.3184,
      "step": 428,
      "task_loss": 0.19590842723846436
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7144808684776381,
      "compression_loss": 0.0,
      "distillation_loss": 0.11836206912994385,
      "epoch": 0.41,
      "learning_rate": 4.958905592107168e-05,
      "loss": 0.1116,
      "step": 429,
      "task_loss": 0.05044740438461304
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7145129501822638,
      "compression_loss": 0.0,
      "distillation_loss": 0.6588773727416992,
      "epoch": 0.41,
      "learning_rate": 4.958712967590756e-05,
      "loss": 0.6266,
      "step": 430,
      "task_loss": 0.3364851772785187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7145450238624512,
      "compression_loss": 0.0,
      "distillation_loss": 0.3294253945350647,
      "epoch": 0.41,
      "learning_rate": 4.9585198964372925e-05,
      "loss": 0.3151,
      "step": 431,
      "task_loss": 0.1865476369857788
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7145770895192041,
      "compression_loss": 0.0,
      "distillation_loss": 0.41862189769744873,
      "epoch": 0.41,
      "learning_rate": 4.958326378681849e-05,
      "loss": 0.3942,
      "step": 432,
      "task_loss": 0.17390699684619904
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7146091471535259,
      "compression_loss": 0.0,
      "distillation_loss": 0.6988288164138794,
      "epoch": 0.41,
      "learning_rate": 4.958132414359579e-05,
      "loss": 0.6708,
      "step": 433,
      "task_loss": 0.41860806941986084
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7146411967664204,
      "compression_loss": 0.0,
      "distillation_loss": 0.5082677006721497,
      "epoch": 0.41,
      "learning_rate": 4.957938003505718e-05,
      "loss": 0.4901,
      "step": 434,
      "task_loss": 0.3265720009803772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7146732383588912,
      "compression_loss": 0.0,
      "distillation_loss": 0.43779420852661133,
      "epoch": 0.41,
      "learning_rate": 4.957743146155581e-05,
      "loss": 0.4178,
      "step": 435,
      "task_loss": 0.23794850707054138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7147052719319422,
      "compression_loss": 0.0,
      "distillation_loss": 0.33255505561828613,
      "epoch": 0.41,
      "learning_rate": 4.9575478423445655e-05,
      "loss": 0.3063,
      "step": 436,
      "task_loss": 0.07019799202680588
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.714737297486577,
      "compression_loss": 0.0,
      "distillation_loss": 0.18722805380821228,
      "epoch": 0.42,
      "learning_rate": 4.957352092108148e-05,
      "loss": 0.1767,
      "step": 437,
      "task_loss": 0.08177616447210312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7147693150237991,
      "compression_loss": 0.0,
      "distillation_loss": 0.44641682505607605,
      "epoch": 0.42,
      "learning_rate": 4.957155895481889e-05,
      "loss": 0.4228,
      "step": 438,
      "task_loss": 0.2099793255329132
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7148013245446124,
      "compression_loss": 0.0,
      "distillation_loss": 0.6220138072967529,
      "epoch": 0.42,
      "learning_rate": 4.956959252501426e-05,
      "loss": 0.5978,
      "step": 439,
      "task_loss": 0.3797354996204376
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7148333260500205,
      "compression_loss": 0.0,
      "distillation_loss": 0.37998977303504944,
      "epoch": 0.42,
      "learning_rate": 4.956762163202484e-05,
      "loss": 0.3565,
      "step": 440,
      "task_loss": 0.1454060971736908
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7148653195410271,
      "compression_loss": 0.0,
      "distillation_loss": 0.5865918397903442,
      "epoch": 0.42,
      "learning_rate": 4.956564627620862e-05,
      "loss": 0.5546,
      "step": 441,
      "task_loss": 0.2664667069911957
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7148973050186359,
      "compression_loss": 0.0,
      "distillation_loss": 0.36456751823425293,
      "epoch": 0.42,
      "learning_rate": 4.956366645792445e-05,
      "loss": 0.3419,
      "step": 442,
      "task_loss": 0.1381196826696396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7149292824838505,
      "compression_loss": 0.0,
      "distillation_loss": 0.3425045609474182,
      "epoch": 0.42,
      "learning_rate": 4.956168217753197e-05,
      "loss": 0.3388,
      "step": 443,
      "task_loss": 0.3051426112651825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7149612519376747,
      "compression_loss": 0.0,
      "distillation_loss": 0.28074485063552856,
      "epoch": 0.42,
      "learning_rate": 4.955969343539162e-05,
      "loss": 0.2671,
      "step": 444,
      "task_loss": 0.14405837655067444
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.714993213381112,
      "compression_loss": 0.0,
      "distillation_loss": 0.16746632754802704,
      "epoch": 0.42,
      "learning_rate": 4.955770023186469e-05,
      "loss": 0.1645,
      "step": 445,
      "task_loss": 0.1381833851337433
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7150251668151663,
      "compression_loss": 0.0,
      "distillation_loss": 0.3588885962963104,
      "epoch": 0.42,
      "learning_rate": 4.9555702567313235e-05,
      "loss": 0.3396,
      "step": 446,
      "task_loss": 0.16621676087379456
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7150571122408411,
      "compression_loss": 0.0,
      "distillation_loss": 0.26380306482315063,
      "epoch": 0.42,
      "learning_rate": 4.9553700442100146e-05,
      "loss": 0.2656,
      "step": 447,
      "task_loss": 0.28156140446662903
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7150890496591402,
      "compression_loss": 0.0,
      "distillation_loss": 0.5757080912590027,
      "epoch": 0.43,
      "learning_rate": 4.955169385658912e-05,
      "loss": 0.5604,
      "step": 448,
      "task_loss": 0.42307889461517334
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7151209790710672,
      "compression_loss": 0.0,
      "distillation_loss": 0.4960484802722931,
      "epoch": 0.43,
      "learning_rate": 4.954968281114467e-05,
      "loss": 0.4804,
      "step": 449,
      "task_loss": 0.33929193019866943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7151529004776259,
      "compression_loss": 0.0,
      "distillation_loss": 0.6576845645904541,
      "epoch": 0.43,
      "learning_rate": 4.9547667306132096e-05,
      "loss": 0.6253,
      "step": 450,
      "task_loss": 0.33397895097732544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7151848138798199,
      "compression_loss": 0.0,
      "distillation_loss": 0.30960074067115784,
      "epoch": 0.43,
      "learning_rate": 4.954564734191753e-05,
      "loss": 0.2977,
      "step": 451,
      "task_loss": 0.1903936266899109
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7152167192786528,
      "compression_loss": 0.0,
      "distillation_loss": 0.1117134764790535,
      "epoch": 0.43,
      "learning_rate": 4.9543622918867926e-05,
      "loss": 0.106,
      "step": 452,
      "task_loss": 0.05442768707871437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7152486166751285,
      "compression_loss": 0.0,
      "distillation_loss": 0.34328725934028625,
      "epoch": 0.43,
      "learning_rate": 4.9541594037351e-05,
      "loss": 0.3272,
      "step": 453,
      "task_loss": 0.1819160431623459
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7152805060702505,
      "compression_loss": 0.0,
      "distillation_loss": 0.6085219979286194,
      "epoch": 0.43,
      "learning_rate": 4.953956069773534e-05,
      "loss": 0.5771,
      "step": 454,
      "task_loss": 0.29387348890304565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7153123874650225,
      "compression_loss": 0.0,
      "distillation_loss": 0.2967539429664612,
      "epoch": 0.43,
      "learning_rate": 4.953752290039028e-05,
      "loss": 0.2804,
      "step": 455,
      "task_loss": 0.13335244357585907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7153442608604482,
      "compression_loss": 0.0,
      "distillation_loss": 0.3782120943069458,
      "epoch": 0.43,
      "learning_rate": 4.953548064568602e-05,
      "loss": 0.3664,
      "step": 456,
      "task_loss": 0.2596074044704437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7153761262575314,
      "compression_loss": 0.0,
      "distillation_loss": 0.21655282378196716,
      "epoch": 0.43,
      "learning_rate": 4.953343393399354e-05,
      "loss": 0.2041,
      "step": 457,
      "task_loss": 0.09245874732732773
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7154079836572756,
      "compression_loss": 0.0,
      "distillation_loss": 0.4280490279197693,
      "epoch": 0.43,
      "learning_rate": 4.953138276568462e-05,
      "loss": 0.4155,
      "step": 458,
      "task_loss": 0.3028221130371094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7154398330606846,
      "compression_loss": 0.0,
      "distillation_loss": 0.6100641489028931,
      "epoch": 0.44,
      "learning_rate": 4.952932714113188e-05,
      "loss": 0.5865,
      "step": 459,
      "task_loss": 0.3739990293979645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7154716744687621,
      "compression_loss": 0.0,
      "distillation_loss": 0.43047159910202026,
      "epoch": 0.44,
      "learning_rate": 4.9527267060708734e-05,
      "loss": 0.4066,
      "step": 460,
      "task_loss": 0.1912587583065033
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7155035078825116,
      "compression_loss": 0.0,
      "distillation_loss": 0.5423746109008789,
      "epoch": 0.44,
      "learning_rate": 4.9525202524789397e-05,
      "loss": 0.5169,
      "step": 461,
      "task_loss": 0.28800442814826965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7155353333029371,
      "compression_loss": 0.0,
      "distillation_loss": 0.2084915190935135,
      "epoch": 0.44,
      "learning_rate": 4.952313353374891e-05,
      "loss": 0.1945,
      "step": 462,
      "task_loss": 0.06886687129735947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7155671507310419,
      "compression_loss": 0.0,
      "distillation_loss": 0.36711233854293823,
      "epoch": 0.44,
      "learning_rate": 4.952106008796311e-05,
      "loss": 0.3632,
      "step": 463,
      "task_loss": 0.3277187943458557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.71559896016783,
      "compression_loss": 0.0,
      "distillation_loss": 0.25077369809150696,
      "epoch": 0.44,
      "learning_rate": 4.9518982187808653e-05,
      "loss": 0.2342,
      "step": 464,
      "task_loss": 0.08512859791517258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7156307616143049,
      "compression_loss": 0.0,
      "distillation_loss": 0.3569698929786682,
      "epoch": 0.44,
      "learning_rate": 4.9516899833663e-05,
      "loss": 0.3385,
      "step": 465,
      "task_loss": 0.17190499603748322
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7156625550714704,
      "compression_loss": 0.0,
      "distillation_loss": 0.2773820161819458,
      "epoch": 0.44,
      "learning_rate": 4.9514813025904413e-05,
      "loss": 0.2626,
      "step": 466,
      "task_loss": 0.12949100136756897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.71569434054033,
      "compression_loss": 0.0,
      "distillation_loss": 0.4447813630104065,
      "epoch": 0.44,
      "learning_rate": 4.951272176491197e-05,
      "loss": 0.4288,
      "step": 467,
      "task_loss": 0.28465670347213745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7157261180218876,
      "compression_loss": 0.0,
      "distillation_loss": 0.41461753845214844,
      "epoch": 0.44,
      "learning_rate": 4.951062605106557e-05,
      "loss": 0.4048,
      "step": 468,
      "task_loss": 0.3163371682167053
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7157578875171468,
      "compression_loss": 0.0,
      "distillation_loss": 0.559532642364502,
      "epoch": 0.45,
      "learning_rate": 4.950852588474591e-05,
      "loss": 0.5354,
      "step": 469,
      "task_loss": 0.31830355525016785
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7157896490271112,
      "compression_loss": 0.0,
      "distillation_loss": 0.28237810730934143,
      "epoch": 0.45,
      "learning_rate": 4.9506421266334475e-05,
      "loss": 0.2687,
      "step": 470,
      "task_loss": 0.14537358283996582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7158214025527846,
      "compression_loss": 0.0,
      "distillation_loss": 0.2132379114627838,
      "epoch": 0.45,
      "learning_rate": 4.9504312196213596e-05,
      "loss": 0.2078,
      "step": 471,
      "task_loss": 0.15892483294010162
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7158531480951705,
      "compression_loss": 0.0,
      "distillation_loss": 0.205579936504364,
      "epoch": 0.45,
      "learning_rate": 4.95021986747664e-05,
      "loss": 0.1931,
      "step": 472,
      "task_loss": 0.08082406967878342
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7158848856552729,
      "compression_loss": 0.0,
      "distillation_loss": 0.31995484232902527,
      "epoch": 0.45,
      "learning_rate": 4.9500080702376805e-05,
      "loss": 0.3075,
      "step": 473,
      "task_loss": 0.19546376168727875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7159166152340952,
      "compression_loss": 0.0,
      "distillation_loss": 0.5218741297721863,
      "epoch": 0.45,
      "learning_rate": 4.949795827942956e-05,
      "loss": 0.4946,
      "step": 474,
      "task_loss": 0.24919168651103973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7159483368326413,
      "compression_loss": 0.0,
      "distillation_loss": 0.49504172801971436,
      "epoch": 0.45,
      "learning_rate": 4.9495831406310205e-05,
      "loss": 0.4802,
      "step": 475,
      "task_loss": 0.34659019112586975
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7159800504519146,
      "compression_loss": 0.0,
      "distillation_loss": 0.43946248292922974,
      "epoch": 0.45,
      "learning_rate": 4.94937000834051e-05,
      "loss": 0.4265,
      "step": 476,
      "task_loss": 0.31031233072280884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.716011756092919,
      "compression_loss": 0.0,
      "distillation_loss": 0.3730089068412781,
      "epoch": 0.45,
      "learning_rate": 4.9491564311101426e-05,
      "loss": 0.3571,
      "step": 477,
      "task_loss": 0.2140880525112152
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7160434537566581,
      "compression_loss": 0.0,
      "distillation_loss": 0.22854942083358765,
      "epoch": 0.45,
      "learning_rate": 4.9489424089787125e-05,
      "loss": 0.214,
      "step": 478,
      "task_loss": 0.08312854915857315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7160751434441356,
      "compression_loss": 0.0,
      "distillation_loss": 0.6738406419754028,
      "epoch": 0.45,
      "learning_rate": 4.948727941985101e-05,
      "loss": 0.6537,
      "step": 479,
      "task_loss": 0.4723435044288635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7161068251563553,
      "compression_loss": 0.0,
      "distillation_loss": 0.44496259093284607,
      "epoch": 0.46,
      "learning_rate": 4.948513030168265e-05,
      "loss": 0.4353,
      "step": 480,
      "task_loss": 0.34817105531692505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7161384988943207,
      "compression_loss": 0.0,
      "distillation_loss": 0.5351232290267944,
      "epoch": 0.46,
      "learning_rate": 4.948297673567245e-05,
      "loss": 0.5138,
      "step": 481,
      "task_loss": 0.3218046724796295
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7161701646590355,
      "compression_loss": 0.0,
      "distillation_loss": 0.5943255424499512,
      "epoch": 0.46,
      "learning_rate": 4.948081872221161e-05,
      "loss": 0.5659,
      "step": 482,
      "task_loss": 0.31032171845436096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7162018224515034,
      "compression_loss": 0.0,
      "distillation_loss": 0.482815146446228,
      "epoch": 0.46,
      "learning_rate": 4.9478656261692155e-05,
      "loss": 0.4567,
      "step": 483,
      "task_loss": 0.22122564911842346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7162334722727283,
      "compression_loss": 0.0,
      "distillation_loss": 0.695792019367218,
      "epoch": 0.46,
      "learning_rate": 4.947648935450689e-05,
      "loss": 0.6668,
      "step": 484,
      "task_loss": 0.40558087825775146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7162651141237135,
      "compression_loss": 0.0,
      "distillation_loss": 0.2511596977710724,
      "epoch": 0.46,
      "learning_rate": 4.947431800104947e-05,
      "loss": 0.2414,
      "step": 485,
      "task_loss": 0.15338845551013947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7162967480054631,
      "compression_loss": 0.0,
      "distillation_loss": 0.45596519112586975,
      "epoch": 0.46,
      "learning_rate": 4.94721422017143e-05,
      "loss": 0.4342,
      "step": 486,
      "task_loss": 0.2383052110671997
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7163283739189804,
      "compression_loss": 0.0,
      "distillation_loss": 0.3888728618621826,
      "epoch": 0.46,
      "learning_rate": 4.946996195689665e-05,
      "loss": 0.3687,
      "step": 487,
      "task_loss": 0.18734854459762573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7163599918652693,
      "compression_loss": 0.0,
      "distillation_loss": 0.47673919796943665,
      "epoch": 0.46,
      "learning_rate": 4.9467777266992555e-05,
      "loss": 0.4547,
      "step": 488,
      "task_loss": 0.2561192512512207
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7163916018453333,
      "compression_loss": 0.0,
      "distillation_loss": 0.3363496959209442,
      "epoch": 0.46,
      "learning_rate": 4.946558813239888e-05,
      "loss": 0.3277,
      "step": 489,
      "task_loss": 0.24951259791851044
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7164232038601764,
      "compression_loss": 0.0,
      "distillation_loss": 0.3862308859825134,
      "epoch": 0.47,
      "learning_rate": 4.94633945535133e-05,
      "loss": 0.3749,
      "step": 490,
      "task_loss": 0.2733776569366455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.716454797910802,
      "compression_loss": 0.0,
      "distillation_loss": 0.2634069323539734,
      "epoch": 0.47,
      "learning_rate": 4.946119653073428e-05,
      "loss": 0.2535,
      "step": 491,
      "task_loss": 0.16414958238601685
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.716486383998214,
      "compression_loss": 0.0,
      "distillation_loss": 0.37143170833587646,
      "epoch": 0.47,
      "learning_rate": 4.9458994064461103e-05,
      "loss": 0.3589,
      "step": 492,
      "task_loss": 0.24603454768657684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7165179621234158,
      "compression_loss": 0.0,
      "distillation_loss": 0.24793553352355957,
      "epoch": 0.47,
      "learning_rate": 4.945678715509386e-05,
      "loss": 0.2349,
      "step": 493,
      "task_loss": 0.11790718138217926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7165495322874113,
      "compression_loss": 0.0,
      "distillation_loss": 0.21886222064495087,
      "epoch": 0.47,
      "learning_rate": 4.9454575803033445e-05,
      "loss": 0.2139,
      "step": 494,
      "task_loss": 0.1692572385072708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7165810944912041,
      "compression_loss": 0.0,
      "distillation_loss": 0.3721332550048828,
      "epoch": 0.47,
      "learning_rate": 4.945236000868156e-05,
      "loss": 0.3528,
      "step": 495,
      "task_loss": 0.17849226295948029
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7166126487357979,
      "compression_loss": 0.0,
      "distillation_loss": 0.4568150043487549,
      "epoch": 0.47,
      "learning_rate": 4.9450139772440715e-05,
      "loss": 0.4353,
      "step": 496,
      "task_loss": 0.24134021997451782
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7166441950221963,
      "compression_loss": 0.0,
      "distillation_loss": 0.20982536673545837,
      "epoch": 0.47,
      "learning_rate": 4.944791509471423e-05,
      "loss": 0.204,
      "step": 497,
      "task_loss": 0.1519790142774582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7166757333514032,
      "compression_loss": 0.0,
      "distillation_loss": 0.30614525079727173,
      "epoch": 0.47,
      "learning_rate": 4.944568597590622e-05,
      "loss": 0.288,
      "step": 498,
      "task_loss": 0.12448100000619888
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7167072637244222,
      "compression_loss": 0.0,
      "distillation_loss": 0.4257122874259949,
      "epoch": 0.47,
      "learning_rate": 4.944345241642162e-05,
      "loss": 0.415,
      "step": 499,
      "task_loss": 0.3188416361808777
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7167387861422567,
      "compression_loss": 0.0,
      "distillation_loss": 0.23117899894714355,
      "epoch": 0.47,
      "learning_rate": 4.944121441666617e-05,
      "loss": 0.2231,
      "step": 500,
      "task_loss": 0.15056748688220978
    },
    {
      "epoch": 0.47,
      "eval_accuracy": 0.8956422018348624,
      "eval_loss": 0.38681745529174805,
      "eval_runtime": 15.2332,
      "eval_samples_per_second": 57.243,
      "eval_steps_per_second": 7.155,
      "step": 500
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7167703006059108,
      "compression_loss": 0.0,
      "distillation_loss": 0.20744021236896515,
      "epoch": 0.48,
      "learning_rate": 4.943897197704642e-05,
      "loss": 0.1999,
      "step": 501,
      "task_loss": 0.13241733610630035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.716801807116388,
      "compression_loss": 0.0,
      "distillation_loss": 0.7784759402275085,
      "epoch": 0.48,
      "learning_rate": 4.9436725097969696e-05,
      "loss": 0.7514,
      "step": 502,
      "task_loss": 0.5077430605888367
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7168333056746918,
      "compression_loss": 0.0,
      "distillation_loss": 0.5257222652435303,
      "epoch": 0.48,
      "learning_rate": 4.943447377984418e-05,
      "loss": 0.5057,
      "step": 503,
      "task_loss": 0.32530543208122253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7168647962818263,
      "compression_loss": 0.0,
      "distillation_loss": 0.13662202656269073,
      "epoch": 0.48,
      "learning_rate": 4.943221802307882e-05,
      "loss": 0.1346,
      "step": 504,
      "task_loss": 0.11593800038099289
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7168962789387948,
      "compression_loss": 0.0,
      "distillation_loss": 0.3498018980026245,
      "epoch": 0.48,
      "learning_rate": 4.942995782808339e-05,
      "loss": 0.3305,
      "step": 505,
      "task_loss": 0.15636515617370605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7169277536466011,
      "compression_loss": 0.0,
      "distillation_loss": 0.3136255145072937,
      "epoch": 0.48,
      "learning_rate": 4.9427693195268466e-05,
      "loss": 0.3083,
      "step": 506,
      "task_loss": 0.26005005836486816
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7169592204062489,
      "compression_loss": 0.0,
      "distillation_loss": 0.27342021465301514,
      "epoch": 0.48,
      "learning_rate": 4.942542412504543e-05,
      "loss": 0.2671,
      "step": 507,
      "task_loss": 0.20997658371925354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.716990679218742,
      "compression_loss": 0.0,
      "distillation_loss": 0.19145192205905914,
      "epoch": 0.48,
      "learning_rate": 4.942315061782646e-05,
      "loss": 0.1941,
      "step": 508,
      "task_loss": 0.21748016774654388
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.717022130085084,
      "compression_loss": 0.0,
      "distillation_loss": 0.21470308303833008,
      "epoch": 0.48,
      "learning_rate": 4.942087267402457e-05,
      "loss": 0.2198,
      "step": 509,
      "task_loss": 0.26588204503059387
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7170535730062784,
      "compression_loss": 0.0,
      "distillation_loss": 0.3804921507835388,
      "epoch": 0.48,
      "learning_rate": 4.941859029405353e-05,
      "loss": 0.3673,
      "step": 510,
      "task_loss": 0.24898777902126312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7170850079833291,
      "compression_loss": 0.0,
      "distillation_loss": 0.3862866759300232,
      "epoch": 0.49,
      "learning_rate": 4.9416303478327974e-05,
      "loss": 0.3702,
      "step": 511,
      "task_loss": 0.22553594410419464
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7171164350172397,
      "compression_loss": 0.0,
      "distillation_loss": 0.10008936375379562,
      "epoch": 0.49,
      "learning_rate": 4.9414012227263295e-05,
      "loss": 0.0956,
      "step": 512,
      "task_loss": 0.05475004017353058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7171478541090139,
      "compression_loss": 0.0,
      "distillation_loss": 0.30363941192626953,
      "epoch": 0.49,
      "learning_rate": 4.941171654127572e-05,
      "loss": 0.2881,
      "step": 513,
      "task_loss": 0.14781013131141663
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7171792652596554,
      "compression_loss": 0.0,
      "distillation_loss": 0.5057163238525391,
      "epoch": 0.49,
      "learning_rate": 4.9409416420782264e-05,
      "loss": 0.4856,
      "step": 514,
      "task_loss": 0.3049323260784149
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7172106684701678,
      "compression_loss": 0.0,
      "distillation_loss": 0.1923336535692215,
      "epoch": 0.49,
      "learning_rate": 4.940711186620076e-05,
      "loss": 0.1797,
      "step": 515,
      "task_loss": 0.06598946452140808
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7172420637415549,
      "compression_loss": 0.0,
      "distillation_loss": 0.5293185710906982,
      "epoch": 0.49,
      "learning_rate": 4.9404802877949843e-05,
      "loss": 0.5265,
      "step": 516,
      "task_loss": 0.5010278820991516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7172734510748203,
      "compression_loss": 0.0,
      "distillation_loss": 0.24947986006736755,
      "epoch": 0.49,
      "learning_rate": 4.940248945644894e-05,
      "loss": 0.2324,
      "step": 517,
      "task_loss": 0.07890691608190536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7173048304709678,
      "compression_loss": 0.0,
      "distillation_loss": 0.27918702363967896,
      "epoch": 0.49,
      "learning_rate": 4.9400171602118306e-05,
      "loss": 0.2666,
      "step": 518,
      "task_loss": 0.15293952822685242
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7173362019310009,
      "compression_loss": 0.0,
      "distillation_loss": 0.4333661198616028,
      "epoch": 0.49,
      "learning_rate": 4.939784931537899e-05,
      "loss": 0.4191,
      "step": 519,
      "task_loss": 0.29058921337127686
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7173675654559234,
      "compression_loss": 0.0,
      "distillation_loss": 0.27411600947380066,
      "epoch": 0.49,
      "learning_rate": 4.9395522596652846e-05,
      "loss": 0.2661,
      "step": 520,
      "task_loss": 0.1940668672323227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.717398921046739,
      "compression_loss": 0.0,
      "distillation_loss": 0.16018861532211304,
      "epoch": 0.49,
      "learning_rate": 4.939319144636253e-05,
      "loss": 0.1733,
      "step": 521,
      "task_loss": 0.2914007306098938
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7174302687044513,
      "compression_loss": 0.0,
      "distillation_loss": 0.2684335708618164,
      "epoch": 0.5,
      "learning_rate": 4.9390855864931504e-05,
      "loss": 0.2554,
      "step": 522,
      "task_loss": 0.13780063390731812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7174616084300639,
      "compression_loss": 0.0,
      "distillation_loss": 0.20900654792785645,
      "epoch": 0.5,
      "learning_rate": 4.938851585278405e-05,
      "loss": 0.1965,
      "step": 523,
      "task_loss": 0.08378442376852036
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7174929402245807,
      "compression_loss": 0.0,
      "distillation_loss": 0.2784028649330139,
      "epoch": 0.5,
      "learning_rate": 4.938617141034523e-05,
      "loss": 0.2624,
      "step": 524,
      "task_loss": 0.11791129410266876
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7175242640890053,
      "compression_loss": 0.0,
      "distillation_loss": 0.12981046736240387,
      "epoch": 0.5,
      "learning_rate": 4.938382253804094e-05,
      "loss": 0.1356,
      "step": 525,
      "task_loss": 0.18778133392333984
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7175555800243414,
      "compression_loss": 0.0,
      "distillation_loss": 0.4506103992462158,
      "epoch": 0.5,
      "learning_rate": 4.938146923629784e-05,
      "loss": 0.4333,
      "step": 526,
      "task_loss": 0.2775367796421051
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7175868880315925,
      "compression_loss": 0.0,
      "distillation_loss": 0.3384746313095093,
      "epoch": 0.5,
      "learning_rate": 4.937911150554343e-05,
      "loss": 0.3378,
      "step": 527,
      "task_loss": 0.3314347267150879
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7176181881117626,
      "compression_loss": 0.0,
      "distillation_loss": 0.2652072608470917,
      "epoch": 0.5,
      "learning_rate": 4.9376749346206006e-05,
      "loss": 0.2549,
      "step": 528,
      "task_loss": 0.16171765327453613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7176494802658551,
      "compression_loss": 0.0,
      "distillation_loss": 0.1128808930516243,
      "epoch": 0.5,
      "learning_rate": 4.937438275871467e-05,
      "loss": 0.1051,
      "step": 529,
      "task_loss": 0.034815624356269836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7176807644948738,
      "compression_loss": 0.0,
      "distillation_loss": 0.335213840007782,
      "epoch": 0.5,
      "learning_rate": 4.9372011743499315e-05,
      "loss": 0.3248,
      "step": 530,
      "task_loss": 0.23120583593845367
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7177120407998224,
      "compression_loss": 0.0,
      "distillation_loss": 0.3072986900806427,
      "epoch": 0.5,
      "learning_rate": 4.9369636300990645e-05,
      "loss": 0.3177,
      "step": 531,
      "task_loss": 0.41148343682289124
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7177433091817046,
      "compression_loss": 0.0,
      "distillation_loss": 0.3255481421947479,
      "epoch": 0.51,
      "learning_rate": 4.936725643162018e-05,
      "loss": 0.3308,
      "step": 532,
      "task_loss": 0.37787866592407227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.717774569641524,
      "compression_loss": 0.0,
      "distillation_loss": 0.8519558906555176,
      "epoch": 0.51,
      "learning_rate": 4.936487213582023e-05,
      "loss": 0.8174,
      "step": 533,
      "task_loss": 0.5065959095954895
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7178058221802843,
      "compression_loss": 0.0,
      "distillation_loss": 0.20490780472755432,
      "epoch": 0.51,
      "learning_rate": 4.9362483414023905e-05,
      "loss": 0.2188,
      "step": 534,
      "task_loss": 0.34347814321517944
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7178370667989893,
      "compression_loss": 0.0,
      "distillation_loss": 0.2214491367340088,
      "epoch": 0.51,
      "learning_rate": 4.936009026666515e-05,
      "loss": 0.2179,
      "step": 535,
      "task_loss": 0.18594780564308167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7178683034986425,
      "compression_loss": 0.0,
      "distillation_loss": 0.2852451801300049,
      "epoch": 0.51,
      "learning_rate": 4.935769269417867e-05,
      "loss": 0.2706,
      "step": 536,
      "task_loss": 0.13881553709506989
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7178995322802476,
      "compression_loss": 0.0,
      "distillation_loss": 0.16195335984230042,
      "epoch": 0.51,
      "learning_rate": 4.935529069700001e-05,
      "loss": 0.1512,
      "step": 537,
      "task_loss": 0.05430576950311661
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7179307531448085,
      "compression_loss": 0.0,
      "distillation_loss": 0.2246689647436142,
      "epoch": 0.51,
      "learning_rate": 4.935288427556549e-05,
      "loss": 0.2136,
      "step": 538,
      "task_loss": 0.11377856135368347
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7179619660933286,
      "compression_loss": 0.0,
      "distillation_loss": 0.17521658539772034,
      "epoch": 0.51,
      "learning_rate": 4.935047343031227e-05,
      "loss": 0.1673,
      "step": 539,
      "task_loss": 0.0958407074213028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7179931711268118,
      "compression_loss": 0.0,
      "distillation_loss": 0.4904990792274475,
      "epoch": 0.51,
      "learning_rate": 4.934805816167827e-05,
      "loss": 0.4757,
      "step": 540,
      "task_loss": 0.3428148627281189
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7180243682462617,
      "compression_loss": 0.0,
      "distillation_loss": 0.362811416387558,
      "epoch": 0.51,
      "learning_rate": 4.934563847010224e-05,
      "loss": 0.3665,
      "step": 541,
      "task_loss": 0.39953047037124634
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7180555574526819,
      "compression_loss": 0.0,
      "distillation_loss": 0.2322240173816681,
      "epoch": 0.51,
      "learning_rate": 4.934321435602374e-05,
      "loss": 0.2193,
      "step": 542,
      "task_loss": 0.10299921035766602
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7180867387470763,
      "compression_loss": 0.0,
      "distillation_loss": 0.15493597090244293,
      "epoch": 0.52,
      "learning_rate": 4.934078581988311e-05,
      "loss": 0.1526,
      "step": 543,
      "task_loss": 0.13205915689468384
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7181179121304484,
      "compression_loss": 0.0,
      "distillation_loss": 0.4433606266975403,
      "epoch": 0.52,
      "learning_rate": 4.933835286212151e-05,
      "loss": 0.4219,
      "step": 544,
      "task_loss": 0.22923287749290466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7181490776038018,
      "compression_loss": 0.0,
      "distillation_loss": 0.3247934579849243,
      "epoch": 0.52,
      "learning_rate": 4.9335915483180896e-05,
      "loss": 0.3249,
      "step": 545,
      "task_loss": 0.3255178928375244
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7181802351681404,
      "compression_loss": 0.0,
      "distillation_loss": 0.6595593690872192,
      "epoch": 0.52,
      "learning_rate": 4.9333473683504025e-05,
      "loss": 0.6289,
      "step": 546,
      "task_loss": 0.35308799147605896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7182113848244678,
      "compression_loss": 0.0,
      "distillation_loss": 0.3609609603881836,
      "epoch": 0.52,
      "learning_rate": 4.9331027463534484e-05,
      "loss": 0.3433,
      "step": 547,
      "task_loss": 0.18484395742416382
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7182425265737876,
      "compression_loss": 0.0,
      "distillation_loss": 0.4003904163837433,
      "epoch": 0.52,
      "learning_rate": 4.932857682371661e-05,
      "loss": 0.3825,
      "step": 548,
      "task_loss": 0.22114846110343933
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7182736604171036,
      "compression_loss": 0.0,
      "distillation_loss": 0.3539576828479767,
      "epoch": 0.52,
      "learning_rate": 4.9326121764495596e-05,
      "loss": 0.3378,
      "step": 549,
      "task_loss": 0.19238103926181793
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7183047863554195,
      "compression_loss": 0.0,
      "distillation_loss": 0.09516200423240662,
      "epoch": 0.52,
      "learning_rate": 4.932366228631741e-05,
      "loss": 0.0975,
      "step": 550,
      "task_loss": 0.11850506067276001
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7183359043897388,
      "compression_loss": 0.0,
      "distillation_loss": 0.784862220287323,
      "epoch": 0.52,
      "learning_rate": 4.932119838962882e-05,
      "loss": 0.7474,
      "step": 551,
      "task_loss": 0.41059941053390503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7183670145210653,
      "compression_loss": 0.0,
      "distillation_loss": 0.3884808421134949,
      "epoch": 0.52,
      "learning_rate": 4.931873007487741e-05,
      "loss": 0.3695,
      "step": 552,
      "task_loss": 0.19908683001995087
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7183981167504028,
      "compression_loss": 0.0,
      "distillation_loss": 0.1411825716495514,
      "epoch": 0.53,
      "learning_rate": 4.9316257342511565e-05,
      "loss": 0.1397,
      "step": 553,
      "task_loss": 0.12602877616882324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7184292110787548,
      "compression_loss": 0.0,
      "distillation_loss": 0.27567747235298157,
      "epoch": 0.53,
      "learning_rate": 4.9313780192980466e-05,
      "loss": 0.259,
      "step": 554,
      "task_loss": 0.10907714068889618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.718460297507125,
      "compression_loss": 0.0,
      "distillation_loss": 0.23973673582077026,
      "epoch": 0.53,
      "learning_rate": 4.9311298626734095e-05,
      "loss": 0.2289,
      "step": 555,
      "task_loss": 0.13170108199119568
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7184913760365172,
      "compression_loss": 0.0,
      "distillation_loss": 0.3157758116722107,
      "epoch": 0.53,
      "learning_rate": 4.9308812644223245e-05,
      "loss": 0.2988,
      "step": 556,
      "task_loss": 0.14566126465797424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7185224466679351,
      "compression_loss": 0.0,
      "distillation_loss": 0.19598382711410522,
      "epoch": 0.53,
      "learning_rate": 4.9306322245899505e-05,
      "loss": 0.187,
      "step": 557,
      "task_loss": 0.10587802529335022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.718553509402382,
      "compression_loss": 0.0,
      "distillation_loss": 0.26901909708976746,
      "epoch": 0.53,
      "learning_rate": 4.930382743221528e-05,
      "loss": 0.2577,
      "step": 558,
      "task_loss": 0.1556887924671173
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7185845642408621,
      "compression_loss": 0.0,
      "distillation_loss": 0.4021989703178406,
      "epoch": 0.53,
      "learning_rate": 4.930132820362374e-05,
      "loss": 0.3864,
      "step": 559,
      "task_loss": 0.24379751086235046
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7186156111843789,
      "compression_loss": 0.0,
      "distillation_loss": 0.4040166735649109,
      "epoch": 0.53,
      "learning_rate": 4.9298824560578895e-05,
      "loss": 0.387,
      "step": 560,
      "task_loss": 0.2337295562028885
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7186466502339359,
      "compression_loss": 0.0,
      "distillation_loss": 0.5355952978134155,
      "epoch": 0.53,
      "learning_rate": 4.929631650353555e-05,
      "loss": 0.5101,
      "step": 561,
      "task_loss": 0.28034472465515137
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.718677681390537,
      "compression_loss": 0.0,
      "distillation_loss": 0.22372055053710938,
      "epoch": 0.53,
      "learning_rate": 4.92938040329493e-05,
      "loss": 0.2111,
      "step": 562,
      "task_loss": 0.09713706374168396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7187087046551858,
      "compression_loss": 0.0,
      "distillation_loss": 0.5858652591705322,
      "epoch": 0.53,
      "learning_rate": 4.9291287149276544e-05,
      "loss": 0.5583,
      "step": 563,
      "task_loss": 0.31010231375694275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7187397200288859,
      "compression_loss": 0.0,
      "distillation_loss": 0.20850016176700592,
      "epoch": 0.54,
      "learning_rate": 4.928876585297448e-05,
      "loss": 0.2016,
      "step": 564,
      "task_loss": 0.1396273374557495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7187707275126411,
      "compression_loss": 0.0,
      "distillation_loss": 0.2773027718067169,
      "epoch": 0.54,
      "learning_rate": 4.9286240144501136e-05,
      "loss": 0.262,
      "step": 565,
      "task_loss": 0.1242566779255867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.718801727107455,
      "compression_loss": 0.0,
      "distillation_loss": 0.11861973255872726,
      "epoch": 0.54,
      "learning_rate": 4.928371002431531e-05,
      "loss": 0.1102,
      "step": 566,
      "task_loss": 0.03490396589040756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7188327188143315,
      "compression_loss": 0.0,
      "distillation_loss": 0.04367408901453018,
      "epoch": 0.54,
      "learning_rate": 4.92811754928766e-05,
      "loss": 0.0404,
      "step": 567,
      "task_loss": 0.010761696845293045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.718863702634274,
      "compression_loss": 0.0,
      "distillation_loss": 0.5102982521057129,
      "epoch": 0.54,
      "learning_rate": 4.927863655064542e-05,
      "loss": 0.4859,
      "step": 568,
      "task_loss": 0.2663833200931549
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7188946785682863,
      "compression_loss": 0.0,
      "distillation_loss": 0.5678519010543823,
      "epoch": 0.54,
      "learning_rate": 4.9276093198082986e-05,
      "loss": 0.5548,
      "step": 569,
      "task_loss": 0.4374026358127594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7189256466173721,
      "compression_loss": 0.0,
      "distillation_loss": 0.2537938952445984,
      "epoch": 0.54,
      "learning_rate": 4.92735454356513e-05,
      "loss": 0.2444,
      "step": 570,
      "task_loss": 0.15972843766212463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.718956606782535,
      "compression_loss": 0.0,
      "distillation_loss": 0.37457334995269775,
      "epoch": 0.54,
      "learning_rate": 4.927099326381319e-05,
      "loss": 0.3575,
      "step": 571,
      "task_loss": 0.20346179604530334
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7189875590647787,
      "compression_loss": 0.0,
      "distillation_loss": 0.4022772014141083,
      "epoch": 0.54,
      "learning_rate": 4.926843668303227e-05,
      "loss": 0.3899,
      "step": 572,
      "task_loss": 0.27801287174224854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7190185034651071,
      "compression_loss": 0.0,
      "distillation_loss": 0.21378515660762787,
      "epoch": 0.54,
      "learning_rate": 4.926587569377293e-05,
      "loss": 0.202,
      "step": 573,
      "task_loss": 0.09593548625707626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7190494399845236,
      "compression_loss": 0.0,
      "distillation_loss": 0.2173890620470047,
      "epoch": 0.55,
      "learning_rate": 4.926331029650042e-05,
      "loss": 0.2096,
      "step": 574,
      "task_loss": 0.1397651731967926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.719080368624032,
      "compression_loss": 0.0,
      "distillation_loss": 0.0907929316163063,
      "epoch": 0.55,
      "learning_rate": 4.926074049168074e-05,
      "loss": 0.0843,
      "step": 575,
      "task_loss": 0.026197172701358795
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.719111289384636,
      "compression_loss": 0.0,
      "distillation_loss": 0.281998872756958,
      "epoch": 0.55,
      "learning_rate": 4.9258166279780704e-05,
      "loss": 0.2744,
      "step": 576,
      "task_loss": 0.20600730180740356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7191422022673393,
      "compression_loss": 0.0,
      "distillation_loss": 0.30183589458465576,
      "epoch": 0.55,
      "learning_rate": 4.925558766126794e-05,
      "loss": 0.2886,
      "step": 577,
      "task_loss": 0.16912756860256195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7191731072731453,
      "compression_loss": 0.0,
      "distillation_loss": 0.49650150537490845,
      "epoch": 0.55,
      "learning_rate": 4.9253004636610856e-05,
      "loss": 0.4803,
      "step": 578,
      "task_loss": 0.33414289355278015
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7192040044030581,
      "compression_loss": 0.0,
      "distillation_loss": 0.39063113927841187,
      "epoch": 0.55,
      "learning_rate": 4.925041720627868e-05,
      "loss": 0.3818,
      "step": 579,
      "task_loss": 0.301964670419693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7192348936580812,
      "compression_loss": 0.0,
      "distillation_loss": 0.34917062520980835,
      "epoch": 0.55,
      "learning_rate": 4.9247825370741416e-05,
      "loss": 0.336,
      "step": 580,
      "task_loss": 0.2170882225036621
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7192657750392183,
      "compression_loss": 0.0,
      "distillation_loss": 0.17031094431877136,
      "epoch": 0.55,
      "learning_rate": 4.924522913046991e-05,
      "loss": 0.1716,
      "step": 581,
      "task_loss": 0.18353326618671417
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.719296648547473,
      "compression_loss": 0.0,
      "distillation_loss": 0.10322088748216629,
      "epoch": 0.55,
      "learning_rate": 4.924262848593576e-05,
      "loss": 0.0994,
      "step": 582,
      "task_loss": 0.0652477964758873
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7193275141838491,
      "compression_loss": 0.0,
      "distillation_loss": 0.11138247698545456,
      "epoch": 0.55,
      "learning_rate": 4.924002343761139e-05,
      "loss": 0.1033,
      "step": 583,
      "task_loss": 0.030367694795131683
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7193583719493502,
      "compression_loss": 0.0,
      "distillation_loss": 0.8268933892250061,
      "epoch": 0.55,
      "learning_rate": 4.923741398597002e-05,
      "loss": 0.7975,
      "step": 584,
      "task_loss": 0.5331718921661377
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7193892218449799,
      "compression_loss": 0.0,
      "distillation_loss": 0.535077691078186,
      "epoch": 0.56,
      "learning_rate": 4.9234800131485675e-05,
      "loss": 0.5079,
      "step": 585,
      "task_loss": 0.263676255941391
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7194200638717421,
      "compression_loss": 0.0,
      "distillation_loss": 0.2574825882911682,
      "epoch": 0.56,
      "learning_rate": 4.9232181874633164e-05,
      "loss": 0.2479,
      "step": 586,
      "task_loss": 0.16171754896640778
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7194508980306403,
      "compression_loss": 0.0,
      "distillation_loss": 0.28252261877059937,
      "epoch": 0.56,
      "learning_rate": 4.922955921588812e-05,
      "loss": 0.2685,
      "step": 587,
      "task_loss": 0.14232273399829865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7194817243226783,
      "compression_loss": 0.0,
      "distillation_loss": 0.5040257573127747,
      "epoch": 0.56,
      "learning_rate": 4.922693215572695e-05,
      "loss": 0.4797,
      "step": 588,
      "task_loss": 0.2605317533016205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7195125427488598,
      "compression_loss": 0.0,
      "distillation_loss": 0.2037096619606018,
      "epoch": 0.56,
      "learning_rate": 4.922430069462688e-05,
      "loss": 0.1908,
      "step": 589,
      "task_loss": 0.07464735209941864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7195433533101884,
      "compression_loss": 0.0,
      "distillation_loss": 0.21558362245559692,
      "epoch": 0.56,
      "learning_rate": 4.9221664833065914e-05,
      "loss": 0.2088,
      "step": 590,
      "task_loss": 0.14768168330192566
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7195741560076677,
      "compression_loss": 0.0,
      "distillation_loss": 0.40281903743743896,
      "epoch": 0.56,
      "learning_rate": 4.921902457152289e-05,
      "loss": 0.3805,
      "step": 591,
      "task_loss": 0.18012678623199463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7196049508423016,
      "compression_loss": 0.0,
      "distillation_loss": 0.3910696804523468,
      "epoch": 0.56,
      "learning_rate": 4.9216379910477403e-05,
      "loss": 0.3727,
      "step": 592,
      "task_loss": 0.20783747732639313
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7196357378150936,
      "compression_loss": 0.0,
      "distillation_loss": 0.24493408203125,
      "epoch": 0.56,
      "learning_rate": 4.921373085040988e-05,
      "loss": 0.2339,
      "step": 593,
      "task_loss": 0.13446438312530518
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7196665169270474,
      "compression_loss": 0.0,
      "distillation_loss": 0.13932454586029053,
      "epoch": 0.56,
      "learning_rate": 4.921107739180153e-05,
      "loss": 0.1317,
      "step": 594,
      "task_loss": 0.06355321407318115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7196972881791668,
      "compression_loss": 0.0,
      "distillation_loss": 0.395768940448761,
      "epoch": 0.57,
      "learning_rate": 4.9208419535134376e-05,
      "loss": 0.384,
      "step": 595,
      "task_loss": 0.2776716947555542
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7197280515724555,
      "compression_loss": 0.0,
      "distillation_loss": 0.24235188961029053,
      "epoch": 0.57,
      "learning_rate": 4.920575728089122e-05,
      "loss": 0.2304,
      "step": 596,
      "task_loss": 0.1224876269698143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.719758807107917,
      "compression_loss": 0.0,
      "distillation_loss": 0.11212563514709473,
      "epoch": 0.57,
      "learning_rate": 4.920309062955568e-05,
      "loss": 0.1317,
      "step": 597,
      "task_loss": 0.3079983592033386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.719789554786555,
      "compression_loss": 0.0,
      "distillation_loss": 0.3031010627746582,
      "epoch": 0.57,
      "learning_rate": 4.920041958161217e-05,
      "loss": 0.2929,
      "step": 598,
      "task_loss": 0.20059747993946075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7198202946093734,
      "compression_loss": 0.0,
      "distillation_loss": 0.6109755039215088,
      "epoch": 0.57,
      "learning_rate": 4.9197744137545884e-05,
      "loss": 0.5924,
      "step": 599,
      "task_loss": 0.4249090254306793
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7198510265773757,
      "compression_loss": 0.0,
      "distillation_loss": 0.33642804622650146,
      "epoch": 0.57,
      "learning_rate": 4.919506429784284e-05,
      "loss": 0.3205,
      "step": 600,
      "task_loss": 0.17708787322044373
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7198817506915656,
      "compression_loss": 0.0,
      "distillation_loss": 0.191384956240654,
      "epoch": 0.57,
      "learning_rate": 4.919238006298984e-05,
      "loss": 0.2127,
      "step": 601,
      "task_loss": 0.40497034788131714
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7199124669529468,
      "compression_loss": 0.0,
      "distillation_loss": 0.19533386826515198,
      "epoch": 0.57,
      "learning_rate": 4.9189691433474494e-05,
      "loss": 0.1888,
      "step": 602,
      "task_loss": 0.12987524271011353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.719943175362523,
      "compression_loss": 0.0,
      "distillation_loss": 0.3835797607898712,
      "epoch": 0.57,
      "learning_rate": 4.91869984097852e-05,
      "loss": 0.3607,
      "step": 603,
      "task_loss": 0.15473908185958862
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7199738759212977,
      "compression_loss": 0.0,
      "distillation_loss": 0.2762410342693329,
      "epoch": 0.57,
      "learning_rate": 4.918430099241116e-05,
      "loss": 0.2624,
      "step": 604,
      "task_loss": 0.137907475233078
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.720004568630275,
      "compression_loss": 0.0,
      "distillation_loss": 0.11789911985397339,
      "epoch": 0.57,
      "learning_rate": 4.918159918184236e-05,
      "loss": 0.1095,
      "step": 605,
      "task_loss": 0.03381138667464256
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7200352534904582,
      "compression_loss": 0.0,
      "distillation_loss": 0.26301029324531555,
      "epoch": 0.58,
      "learning_rate": 4.9178892978569625e-05,
      "loss": 0.2488,
      "step": 606,
      "task_loss": 0.12070365250110626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7200659305028512,
      "compression_loss": 0.0,
      "distillation_loss": 0.1380884349346161,
      "epoch": 0.58,
      "learning_rate": 4.9176182383084524e-05,
      "loss": 0.1346,
      "step": 607,
      "task_loss": 0.10351494699716568
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7200965996684575,
      "compression_loss": 0.0,
      "distillation_loss": 0.4840831756591797,
      "epoch": 0.58,
      "learning_rate": 4.917346739587946e-05,
      "loss": 0.4609,
      "step": 608,
      "task_loss": 0.25223809480667114
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.720127260988281,
      "compression_loss": 0.0,
      "distillation_loss": 0.28585392236709595,
      "epoch": 0.58,
      "learning_rate": 4.917074801744763e-05,
      "loss": 0.2685,
      "step": 609,
      "task_loss": 0.112055703997612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7201579144633251,
      "compression_loss": 0.0,
      "distillation_loss": 0.379632830619812,
      "epoch": 0.58,
      "learning_rate": 4.916802424828301e-05,
      "loss": 0.3577,
      "step": 610,
      "task_loss": 0.15994824469089508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7201885600945938,
      "compression_loss": 0.0,
      "distillation_loss": 0.3786671757698059,
      "epoch": 0.58,
      "learning_rate": 4.9165296088880384e-05,
      "loss": 0.3794,
      "step": 611,
      "task_loss": 0.38576364517211914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7202191978830905,
      "compression_loss": 0.0,
      "distillation_loss": 0.2917826771736145,
      "epoch": 0.58,
      "learning_rate": 4.916256353973535e-05,
      "loss": 0.2759,
      "step": 612,
      "task_loss": 0.13341909646987915
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7202498278298191,
      "compression_loss": 0.0,
      "distillation_loss": 0.24248436093330383,
      "epoch": 0.58,
      "learning_rate": 4.9159826601344286e-05,
      "loss": 0.2274,
      "step": 613,
      "task_loss": 0.0919809564948082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7202804499357831,
      "compression_loss": 0.0,
      "distillation_loss": 0.12847685813903809,
      "epoch": 0.58,
      "learning_rate": 4.915708527420435e-05,
      "loss": 0.1231,
      "step": 614,
      "task_loss": 0.0744004100561142
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7203110642019863,
      "compression_loss": 0.0,
      "distillation_loss": 0.1666882485151291,
      "epoch": 0.58,
      "learning_rate": 4.9154339558813546e-05,
      "loss": 0.1631,
      "step": 615,
      "task_loss": 0.13053244352340698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7203416706294323,
      "compression_loss": 0.0,
      "distillation_loss": 0.7391297817230225,
      "epoch": 0.58,
      "learning_rate": 4.915158945567062e-05,
      "loss": 0.7177,
      "step": 616,
      "task_loss": 0.5249520540237427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.720372269219125,
      "compression_loss": 0.0,
      "distillation_loss": 0.18314534425735474,
      "epoch": 0.59,
      "learning_rate": 4.914883496527516e-05,
      "loss": 0.1906,
      "step": 617,
      "task_loss": 0.2574879825115204
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7204028599720679,
      "compression_loss": 0.0,
      "distillation_loss": 0.38090646266937256,
      "epoch": 0.59,
      "learning_rate": 4.914607608812753e-05,
      "loss": 0.3593,
      "step": 618,
      "task_loss": 0.1652752161026001
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7204334428892646,
      "compression_loss": 0.0,
      "distillation_loss": 0.1799452304840088,
      "epoch": 0.59,
      "learning_rate": 4.9143312824728896e-05,
      "loss": 0.1885,
      "step": 619,
      "task_loss": 0.26563724875450134
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7204640179717189,
      "compression_loss": 0.0,
      "distillation_loss": 0.2368253916501999,
      "epoch": 0.59,
      "learning_rate": 4.91405451755812e-05,
      "loss": 0.2235,
      "step": 620,
      "task_loss": 0.10326235741376877
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7204945852204344,
      "compression_loss": 0.0,
      "distillation_loss": 0.5095282196998596,
      "epoch": 0.59,
      "learning_rate": 4.913777314118721e-05,
      "loss": 0.4846,
      "step": 621,
      "task_loss": 0.2607031464576721
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7205251446364149,
      "compression_loss": 0.0,
      "distillation_loss": 0.12455812841653824,
      "epoch": 0.59,
      "learning_rate": 4.9134996722050483e-05,
      "loss": 0.1324,
      "step": 622,
      "task_loss": 0.20319408178329468
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7205556962206641,
      "compression_loss": 0.0,
      "distillation_loss": 0.3534661531448364,
      "epoch": 0.59,
      "learning_rate": 4.913221591867537e-05,
      "loss": 0.3352,
      "step": 623,
      "task_loss": 0.17124255001544952
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7205862399741856,
      "compression_loss": 0.0,
      "distillation_loss": 0.06430540978908539,
      "epoch": 0.59,
      "learning_rate": 4.912943073156701e-05,
      "loss": 0.0592,
      "step": 624,
      "task_loss": 0.013450298458337784
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.720616775897983,
      "compression_loss": 0.0,
      "distillation_loss": 0.26247042417526245,
      "epoch": 0.59,
      "learning_rate": 4.912664116123134e-05,
      "loss": 0.2455,
      "step": 625,
      "task_loss": 0.09270986914634705
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7206473039930601,
      "compression_loss": 0.0,
      "distillation_loss": 0.5087012052536011,
      "epoch": 0.59,
      "learning_rate": 4.9123847208175126e-05,
      "loss": 0.4844,
      "step": 626,
      "task_loss": 0.2655639946460724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7206778242604206,
      "compression_loss": 0.0,
      "distillation_loss": 0.2001875340938568,
      "epoch": 0.6,
      "learning_rate": 4.912104887290587e-05,
      "loss": 0.1939,
      "step": 627,
      "task_loss": 0.13689836859703064
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7207083367010682,
      "compression_loss": 0.0,
      "distillation_loss": 0.19963237643241882,
      "epoch": 0.6,
      "learning_rate": 4.911824615593193e-05,
      "loss": 0.1886,
      "step": 628,
      "task_loss": 0.08933916687965393
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7207388413160064,
      "compression_loss": 0.0,
      "distillation_loss": 0.4403138756752014,
      "epoch": 0.6,
      "learning_rate": 4.9115439057762416e-05,
      "loss": 0.4213,
      "step": 629,
      "task_loss": 0.2505990266799927
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.720769338106239,
      "compression_loss": 0.0,
      "distillation_loss": 0.4388275146484375,
      "epoch": 0.6,
      "learning_rate": 4.911262757890726e-05,
      "loss": 0.4296,
      "step": 630,
      "task_loss": 0.3462795615196228
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7207998270727697,
      "compression_loss": 0.0,
      "distillation_loss": 0.33021849393844604,
      "epoch": 0.6,
      "learning_rate": 4.9109811719877166e-05,
      "loss": 0.3132,
      "step": 631,
      "task_loss": 0.16027683019638062
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7208303082166022,
      "compression_loss": 0.0,
      "distillation_loss": 0.483683317899704,
      "epoch": 0.6,
      "learning_rate": 4.910699148118367e-05,
      "loss": 0.4708,
      "step": 632,
      "task_loss": 0.3550935983657837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7208607815387402,
      "compression_loss": 0.0,
      "distillation_loss": 0.39671921730041504,
      "epoch": 0.6,
      "learning_rate": 4.910416686333906e-05,
      "loss": 0.3752,
      "step": 633,
      "task_loss": 0.18194958567619324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7208912470401873,
      "compression_loss": 0.0,
      "distillation_loss": 0.5867143273353577,
      "epoch": 0.6,
      "learning_rate": 4.910133786685646e-05,
      "loss": 0.5733,
      "step": 634,
      "task_loss": 0.4523051083087921
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7209217047219472,
      "compression_loss": 0.0,
      "distillation_loss": 0.43349331617355347,
      "epoch": 0.6,
      "learning_rate": 4.9098504492249764e-05,
      "loss": 0.4114,
      "step": 635,
      "task_loss": 0.2126692235469818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7209521545850235,
      "compression_loss": 0.0,
      "distillation_loss": 0.35138991475105286,
      "epoch": 0.6,
      "learning_rate": 4.9095666740033664e-05,
      "loss": 0.3348,
      "step": 636,
      "task_loss": 0.18542218208312988
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7209825966304201,
      "compression_loss": 0.0,
      "distillation_loss": 0.43535393476486206,
      "epoch": 0.6,
      "learning_rate": 4.9092824610723655e-05,
      "loss": 0.4145,
      "step": 637,
      "task_loss": 0.22690628468990326
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7210130308591405,
      "compression_loss": 0.0,
      "distillation_loss": 0.2631605565547943,
      "epoch": 0.61,
      "learning_rate": 4.908997810483602e-05,
      "loss": 0.2534,
      "step": 638,
      "task_loss": 0.1657785326242447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7210434572721884,
      "compression_loss": 0.0,
      "distillation_loss": 0.4615224003791809,
      "epoch": 0.61,
      "learning_rate": 4.908712722288785e-05,
      "loss": 0.4336,
      "step": 639,
      "task_loss": 0.1823965460062027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7210738758705676,
      "compression_loss": 0.0,
      "distillation_loss": 0.3323667645454407,
      "epoch": 0.61,
      "learning_rate": 4.9084271965397014e-05,
      "loss": 0.3282,
      "step": 640,
      "task_loss": 0.290405809879303
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7211042866552817,
      "compression_loss": 0.0,
      "distillation_loss": 0.27497389912605286,
      "epoch": 0.61,
      "learning_rate": 4.908141233288218e-05,
      "loss": 0.2726,
      "step": 641,
      "task_loss": 0.2515375018119812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7211346896273344,
      "compression_loss": 0.0,
      "distillation_loss": 0.1021830216050148,
      "epoch": 0.61,
      "learning_rate": 4.907854832586282e-05,
      "loss": 0.094,
      "step": 642,
      "task_loss": 0.020755015313625336
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7211650847877293,
      "compression_loss": 0.0,
      "distillation_loss": 0.19680382311344147,
      "epoch": 0.61,
      "learning_rate": 4.907567994485919e-05,
      "loss": 0.1992,
      "step": 643,
      "task_loss": 0.2208843231201172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7211954721374703,
      "compression_loss": 0.0,
      "distillation_loss": 0.5825210809707642,
      "epoch": 0.61,
      "learning_rate": 4.9072807190392354e-05,
      "loss": 0.5626,
      "step": 644,
      "task_loss": 0.383215069770813
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7212258516775608,
      "compression_loss": 0.0,
      "distillation_loss": 0.3526345491409302,
      "epoch": 0.61,
      "learning_rate": 4.906993006298416e-05,
      "loss": 0.3389,
      "step": 645,
      "task_loss": 0.21486197412014008
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7212562234090046,
      "compression_loss": 0.0,
      "distillation_loss": 0.4113069772720337,
      "epoch": 0.61,
      "learning_rate": 4.9067048563157235e-05,
      "loss": 0.4089,
      "step": 646,
      "task_loss": 0.38709861040115356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7212865873328055,
      "compression_loss": 0.0,
      "distillation_loss": 0.27968546748161316,
      "epoch": 0.61,
      "learning_rate": 4.906416269143505e-05,
      "loss": 0.2654,
      "step": 647,
      "task_loss": 0.13687899708747864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.721316943449967,
      "compression_loss": 0.0,
      "distillation_loss": 0.4654969274997711,
      "epoch": 0.62,
      "learning_rate": 4.90612724483418e-05,
      "loss": 0.4424,
      "step": 648,
      "task_loss": 0.23491407930850983
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7213472917614929,
      "compression_loss": 0.0,
      "distillation_loss": 0.35083064436912537,
      "epoch": 0.62,
      "learning_rate": 4.905837783440253e-05,
      "loss": 0.3333,
      "step": 649,
      "task_loss": 0.17527225613594055
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7213776322683869,
      "compression_loss": 0.0,
      "distillation_loss": 0.7668882608413696,
      "epoch": 0.62,
      "learning_rate": 4.905547885014307e-05,
      "loss": 0.7316,
      "step": 650,
      "task_loss": 0.4138070046901703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7214079649716525,
      "compression_loss": 0.0,
      "distillation_loss": 0.25231415033340454,
      "epoch": 0.62,
      "learning_rate": 4.9052575496090016e-05,
      "loss": 0.2521,
      "step": 651,
      "task_loss": 0.2498636394739151
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7214382898722936,
      "compression_loss": 0.0,
      "distillation_loss": 0.39607787132263184,
      "epoch": 0.62,
      "learning_rate": 4.904966777277079e-05,
      "loss": 0.3755,
      "step": 652,
      "task_loss": 0.19066929817199707
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7214686069713138,
      "compression_loss": 0.0,
      "distillation_loss": 0.33866336941719055,
      "epoch": 0.62,
      "learning_rate": 4.9046755680713586e-05,
      "loss": 0.3344,
      "step": 653,
      "task_loss": 0.2955394387245178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7214989162697167,
      "compression_loss": 0.0,
      "distillation_loss": 0.26593947410583496,
      "epoch": 0.62,
      "learning_rate": 4.90438392204474e-05,
      "loss": 0.2494,
      "step": 654,
      "task_loss": 0.10030423104763031
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7215292177685061,
      "compression_loss": 0.0,
      "distillation_loss": 0.4320967495441437,
      "epoch": 0.62,
      "learning_rate": 4.9040918392502026e-05,
      "loss": 0.4192,
      "step": 655,
      "task_loss": 0.30289995670318604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7215595114686856,
      "compression_loss": 0.0,
      "distillation_loss": 0.15407636761665344,
      "epoch": 0.62,
      "learning_rate": 4.903799319740804e-05,
      "loss": 0.145,
      "step": 656,
      "task_loss": 0.0632171630859375
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.721589797371259,
      "compression_loss": 0.0,
      "distillation_loss": 0.11901634931564331,
      "epoch": 0.62,
      "learning_rate": 4.903506363569683e-05,
      "loss": 0.1128,
      "step": 657,
      "task_loss": 0.05649835988879204
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7216200754772298,
      "compression_loss": 0.0,
      "distillation_loss": 0.2876704931259155,
      "epoch": 0.62,
      "learning_rate": 4.9032129707900556e-05,
      "loss": 0.283,
      "step": 658,
      "task_loss": 0.2414519190788269
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7216503457876018,
      "compression_loss": 0.0,
      "distillation_loss": 0.1815006583929062,
      "epoch": 0.63,
      "learning_rate": 4.9029191414552165e-05,
      "loss": 0.1867,
      "step": 659,
      "task_loss": 0.23326468467712402
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7216806083033788,
      "compression_loss": 0.0,
      "distillation_loss": 0.28646424412727356,
      "epoch": 0.63,
      "learning_rate": 4.9026248756185445e-05,
      "loss": 0.2784,
      "step": 660,
      "task_loss": 0.20588558912277222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7217108630255642,
      "compression_loss": 0.0,
      "distillation_loss": 0.22158020734786987,
      "epoch": 0.63,
      "learning_rate": 4.902330173333492e-05,
      "loss": 0.2171,
      "step": 661,
      "task_loss": 0.17726564407348633
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7217411099551618,
      "compression_loss": 0.0,
      "distillation_loss": 0.41758403182029724,
      "epoch": 0.63,
      "learning_rate": 4.9020350346535936e-05,
      "loss": 0.3971,
      "step": 662,
      "task_loss": 0.21297332644462585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7217713490931754,
      "compression_loss": 0.0,
      "distillation_loss": 0.16590705513954163,
      "epoch": 0.63,
      "learning_rate": 4.901739459632463e-05,
      "loss": 0.1547,
      "step": 663,
      "task_loss": 0.05394207686185837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7218015804406086,
      "compression_loss": 0.0,
      "distillation_loss": 0.22256065905094147,
      "epoch": 0.63,
      "learning_rate": 4.901443448323792e-05,
      "loss": 0.2239,
      "step": 664,
      "task_loss": 0.2363419383764267
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.721831803998465,
      "compression_loss": 0.0,
      "distillation_loss": 0.17048153281211853,
      "epoch": 0.63,
      "learning_rate": 4.901147000781355e-05,
      "loss": 0.1603,
      "step": 665,
      "task_loss": 0.068704754114151
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7218620197677484,
      "compression_loss": 0.0,
      "distillation_loss": 0.5146390199661255,
      "epoch": 0.63,
      "learning_rate": 4.9008501170589996e-05,
      "loss": 0.4834,
      "step": 666,
      "task_loss": 0.20194561779499054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7218922277494625,
      "compression_loss": 0.0,
      "distillation_loss": 0.27669647336006165,
      "epoch": 0.63,
      "learning_rate": 4.900552797210658e-05,
      "loss": 0.2729,
      "step": 667,
      "task_loss": 0.23846843838691711
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7219224279446109,
      "compression_loss": 0.0,
      "distillation_loss": 0.5004544258117676,
      "epoch": 0.63,
      "learning_rate": 4.90025504129034e-05,
      "loss": 0.4772,
      "step": 668,
      "task_loss": 0.26774314045906067
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7219526203541973,
      "compression_loss": 0.0,
      "distillation_loss": 0.15412703156471252,
      "epoch": 0.64,
      "learning_rate": 4.8999568493521345e-05,
      "loss": 0.1447,
      "step": 669,
      "task_loss": 0.0594392754137516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7219828049792254,
      "compression_loss": 0.0,
      "distillation_loss": 0.34250378608703613,
      "epoch": 0.64,
      "learning_rate": 4.899658221450208e-05,
      "loss": 0.3341,
      "step": 670,
      "task_loss": 0.2580875754356384
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7220129818206988,
      "compression_loss": 0.0,
      "distillation_loss": 0.334232360124588,
      "epoch": 0.64,
      "learning_rate": 4.899359157638809e-05,
      "loss": 0.3119,
      "step": 671,
      "task_loss": 0.1106051504611969
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7220431508796213,
      "compression_loss": 0.0,
      "distillation_loss": 0.15714368224143982,
      "epoch": 0.64,
      "learning_rate": 4.899059657972264e-05,
      "loss": 0.1483,
      "step": 672,
      "task_loss": 0.06857656687498093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7220733121569965,
      "compression_loss": 0.0,
      "distillation_loss": 0.47153952717781067,
      "epoch": 0.64,
      "learning_rate": 4.898759722504977e-05,
      "loss": 0.4422,
      "step": 673,
      "task_loss": 0.17770282924175262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7221034656538281,
      "compression_loss": 0.0,
      "distillation_loss": 0.24304763972759247,
      "epoch": 0.64,
      "learning_rate": 4.8984593512914356e-05,
      "loss": 0.2322,
      "step": 674,
      "task_loss": 0.1346360594034195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7221336113711199,
      "compression_loss": 0.0,
      "distillation_loss": 0.20216722786426544,
      "epoch": 0.64,
      "learning_rate": 4.898158544386201e-05,
      "loss": 0.2003,
      "step": 675,
      "task_loss": 0.18363113701343536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7221637493098754,
      "compression_loss": 0.0,
      "distillation_loss": 0.3209993243217468,
      "epoch": 0.64,
      "learning_rate": 4.897857301843917e-05,
      "loss": 0.3081,
      "step": 676,
      "task_loss": 0.19210395216941833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7221938794710984,
      "compression_loss": 0.0,
      "distillation_loss": 0.38004785776138306,
      "epoch": 0.64,
      "learning_rate": 4.897555623719306e-05,
      "loss": 0.3761,
      "step": 677,
      "task_loss": 0.34037119150161743
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7222240018557925,
      "compression_loss": 0.0,
      "distillation_loss": 0.27278655767440796,
      "epoch": 0.64,
      "learning_rate": 4.897253510067169e-05,
      "loss": 0.2581,
      "step": 678,
      "task_loss": 0.12568269670009613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7222541164649614,
      "compression_loss": 0.0,
      "distillation_loss": 0.48017337918281555,
      "epoch": 0.64,
      "learning_rate": 4.896950960942387e-05,
      "loss": 0.4528,
      "step": 679,
      "task_loss": 0.20639115571975708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7222842232996088,
      "compression_loss": 0.0,
      "distillation_loss": 0.12991678714752197,
      "epoch": 0.65,
      "learning_rate": 4.896647976399919e-05,
      "loss": 0.1368,
      "step": 680,
      "task_loss": 0.19825144112110138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7223143223607386,
      "compression_loss": 0.0,
      "distillation_loss": 0.2807837724685669,
      "epoch": 0.65,
      "learning_rate": 4.896344556494804e-05,
      "loss": 0.2644,
      "step": 681,
      "task_loss": 0.11680983006954193
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.722344413649354,
      "compression_loss": 0.0,
      "distillation_loss": 0.36504942178726196,
      "epoch": 0.65,
      "learning_rate": 4.8960407012821584e-05,
      "loss": 0.3483,
      "step": 682,
      "task_loss": 0.19777251780033112
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7223744971664591,
      "compression_loss": 0.0,
      "distillation_loss": 0.24195873737335205,
      "epoch": 0.65,
      "learning_rate": 4.895736410817181e-05,
      "loss": 0.2359,
      "step": 683,
      "task_loss": 0.1811816245317459
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7224045729130574,
      "compression_loss": 0.0,
      "distillation_loss": 0.17951583862304688,
      "epoch": 0.65,
      "learning_rate": 4.8954316851551465e-05,
      "loss": 0.1744,
      "step": 684,
      "task_loss": 0.12814725935459137
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7224346408901526,
      "compression_loss": 0.0,
      "distillation_loss": 0.3897627592086792,
      "epoch": 0.65,
      "learning_rate": 4.895126524351409e-05,
      "loss": 0.3718,
      "step": 685,
      "task_loss": 0.20980247855186462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7224647010987484,
      "compression_loss": 0.0,
      "distillation_loss": 0.15180069208145142,
      "epoch": 0.65,
      "learning_rate": 4.8948209284614046e-05,
      "loss": 0.144,
      "step": 686,
      "task_loss": 0.073697030544281
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7224947535398485,
      "compression_loss": 0.0,
      "distillation_loss": 0.3309364914894104,
      "epoch": 0.65,
      "learning_rate": 4.894514897540643e-05,
      "loss": 0.3178,
      "step": 687,
      "task_loss": 0.1999032199382782
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7225247982144566,
      "compression_loss": 0.0,
      "distillation_loss": 0.5048620700836182,
      "epoch": 0.65,
      "learning_rate": 4.89420843164472e-05,
      "loss": 0.475,
      "step": 688,
      "task_loss": 0.20606492459774017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7225548351235763,
      "compression_loss": 0.0,
      "distillation_loss": 0.26741254329681396,
      "epoch": 0.65,
      "learning_rate": 4.893901530829304e-05,
      "loss": 0.2534,
      "step": 689,
      "task_loss": 0.12686991691589355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7225848642682113,
      "compression_loss": 0.0,
      "distillation_loss": 0.454744428396225,
      "epoch": 0.66,
      "learning_rate": 4.8935941951501463e-05,
      "loss": 0.4328,
      "step": 690,
      "task_loss": 0.2351890504360199
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7226148856493654,
      "compression_loss": 0.0,
      "distillation_loss": 0.4148391783237457,
      "epoch": 0.66,
      "learning_rate": 4.893286424663075e-05,
      "loss": 0.391,
      "step": 691,
      "task_loss": 0.17629502713680267
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7226448992680421,
      "compression_loss": 0.0,
      "distillation_loss": 0.3470875322818756,
      "epoch": 0.66,
      "learning_rate": 4.892978219423998e-05,
      "loss": 0.3301,
      "step": 692,
      "task_loss": 0.17764955759048462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7226749051252452,
      "compression_loss": 0.0,
      "distillation_loss": 0.31869742274284363,
      "epoch": 0.66,
      "learning_rate": 4.892669579488903e-05,
      "loss": 0.3036,
      "step": 693,
      "task_loss": 0.16786476969718933
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7227049032219783,
      "compression_loss": 0.0,
      "distillation_loss": 0.24134713411331177,
      "epoch": 0.66,
      "learning_rate": 4.892360504913856e-05,
      "loss": 0.2292,
      "step": 694,
      "task_loss": 0.12003158777952194
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7227348935592453,
      "compression_loss": 0.0,
      "distillation_loss": 0.19437089562416077,
      "epoch": 0.66,
      "learning_rate": 4.8920509957550016e-05,
      "loss": 0.1999,
      "step": 695,
      "task_loss": 0.24951303005218506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7227648761380496,
      "compression_loss": 0.0,
      "distillation_loss": 0.2377103865146637,
      "epoch": 0.66,
      "learning_rate": 4.8917410520685635e-05,
      "loss": 0.2238,
      "step": 696,
      "task_loss": 0.09895863384008408
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7227948509593951,
      "compression_loss": 0.0,
      "distillation_loss": 0.09030158072710037,
      "epoch": 0.66,
      "learning_rate": 4.891430673910844e-05,
      "loss": 0.0912,
      "step": 697,
      "task_loss": 0.09956920146942139
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7228248180242853,
      "compression_loss": 0.0,
      "distillation_loss": 0.20243610441684723,
      "epoch": 0.66,
      "learning_rate": 4.891119861338226e-05,
      "loss": 0.1953,
      "step": 698,
      "task_loss": 0.13071851432323456
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7228547773337239,
      "compression_loss": 0.0,
      "distillation_loss": 0.1336769163608551,
      "epoch": 0.66,
      "learning_rate": 4.8908086144071694e-05,
      "loss": 0.1241,
      "step": 699,
      "task_loss": 0.03775034099817276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7228847288887148,
      "compression_loss": 0.0,
      "distillation_loss": 0.16907915472984314,
      "epoch": 0.66,
      "learning_rate": 4.8904969331742136e-05,
      "loss": 0.1675,
      "step": 700,
      "task_loss": 0.15347930788993835
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7229146726902614,
      "compression_loss": 0.0,
      "distillation_loss": 0.15456578135490417,
      "epoch": 0.67,
      "learning_rate": 4.890184817695976e-05,
      "loss": 0.1522,
      "step": 701,
      "task_loss": 0.13044598698616028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7229446087393676,
      "compression_loss": 0.0,
      "distillation_loss": 0.18475860357284546,
      "epoch": 0.67,
      "learning_rate": 4.8898722680291564e-05,
      "loss": 0.1854,
      "step": 702,
      "task_loss": 0.19159327447414398
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7229745370370371,
      "compression_loss": 0.0,
      "distillation_loss": 0.3150096535682678,
      "epoch": 0.67,
      "learning_rate": 4.8895592842305295e-05,
      "loss": 0.3036,
      "step": 703,
      "task_loss": 0.20053307712078094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7230044575842732,
      "compression_loss": 0.0,
      "distillation_loss": 0.3283127546310425,
      "epoch": 0.67,
      "learning_rate": 4.88924586635695e-05,
      "loss": 0.322,
      "step": 704,
      "task_loss": 0.2651616036891937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7230343703820801,
      "compression_loss": 0.0,
      "distillation_loss": 0.19642552733421326,
      "epoch": 0.67,
      "learning_rate": 4.888932014465352e-05,
      "loss": 0.1945,
      "step": 705,
      "task_loss": 0.1771748960018158
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7230642754314611,
      "compression_loss": 0.0,
      "distillation_loss": 0.23262354731559753,
      "epoch": 0.67,
      "learning_rate": 4.888617728612749e-05,
      "loss": 0.2279,
      "step": 706,
      "task_loss": 0.18583974242210388
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7230941727334201,
      "compression_loss": 0.0,
      "distillation_loss": 0.329001247882843,
      "epoch": 0.67,
      "learning_rate": 4.888303008856231e-05,
      "loss": 0.3181,
      "step": 707,
      "task_loss": 0.21996405720710754
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7231240622889608,
      "compression_loss": 0.0,
      "distillation_loss": 0.5461586713790894,
      "epoch": 0.67,
      "learning_rate": 4.88798785525297e-05,
      "loss": 0.5225,
      "step": 708,
      "task_loss": 0.3095916509628296
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7231539440990866,
      "compression_loss": 0.0,
      "distillation_loss": 0.29736411571502686,
      "epoch": 0.67,
      "learning_rate": 4.887672267860214e-05,
      "loss": 0.2917,
      "step": 709,
      "task_loss": 0.2409740537405014
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7231838181648014,
      "compression_loss": 0.0,
      "distillation_loss": 0.33324652910232544,
      "epoch": 0.67,
      "learning_rate": 4.887356246735292e-05,
      "loss": 0.3262,
      "step": 710,
      "task_loss": 0.2625032365322113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7232136844871089,
      "compression_loss": 0.0,
      "distillation_loss": 0.3226258158683777,
      "epoch": 0.68,
      "learning_rate": 4.8870397919356094e-05,
      "loss": 0.3109,
      "step": 711,
      "task_loss": 0.20536869764328003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7232435430670128,
      "compression_loss": 0.0,
      "distillation_loss": 0.3054695427417755,
      "epoch": 0.68,
      "learning_rate": 4.8867229035186526e-05,
      "loss": 0.3009,
      "step": 712,
      "task_loss": 0.25971150398254395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7232733939055166,
      "compression_loss": 0.0,
      "distillation_loss": 0.2144642472267151,
      "epoch": 0.68,
      "learning_rate": 4.886405581541986e-05,
      "loss": 0.2068,
      "step": 713,
      "task_loss": 0.13790678977966309
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7233032370036241,
      "compression_loss": 0.0,
      "distillation_loss": 0.193600594997406,
      "epoch": 0.68,
      "learning_rate": 4.886087826063252e-05,
      "loss": 0.1853,
      "step": 714,
      "task_loss": 0.11010687053203583
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.723333072362339,
      "compression_loss": 0.0,
      "distillation_loss": 0.3077680170536041,
      "epoch": 0.68,
      "learning_rate": 4.8857696371401735e-05,
      "loss": 0.2948,
      "step": 715,
      "task_loss": 0.1778283566236496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.723362899982665,
      "compression_loss": 0.0,
      "distillation_loss": 0.24543452262878418,
      "epoch": 0.68,
      "learning_rate": 4.88545101483055e-05,
      "loss": 0.2343,
      "step": 716,
      "task_loss": 0.13451889157295227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7233927198656058,
      "compression_loss": 0.0,
      "distillation_loss": 0.4300387501716614,
      "epoch": 0.68,
      "learning_rate": 4.885131959192262e-05,
      "loss": 0.412,
      "step": 717,
      "task_loss": 0.24990172684192657
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7234225320121649,
      "compression_loss": 0.0,
      "distillation_loss": 0.368339866399765,
      "epoch": 0.68,
      "learning_rate": 4.884812470283265e-05,
      "loss": 0.3664,
      "step": 718,
      "task_loss": 0.348560631275177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7234523364233462,
      "compression_loss": 0.0,
      "distillation_loss": 0.18903806805610657,
      "epoch": 0.68,
      "learning_rate": 4.884492548161599e-05,
      "loss": 0.186,
      "step": 719,
      "task_loss": 0.1587422788143158
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7234821331001532,
      "compression_loss": 0.0,
      "distillation_loss": 0.2814609110355377,
      "epoch": 0.68,
      "learning_rate": 4.8841721928853776e-05,
      "loss": 0.2723,
      "step": 720,
      "task_loss": 0.19001665711402893
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7235119220435898,
      "compression_loss": 0.0,
      "distillation_loss": 0.4020082950592041,
      "epoch": 0.68,
      "learning_rate": 4.8838514045127945e-05,
      "loss": 0.385,
      "step": 721,
      "task_loss": 0.23192302882671356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7235417032546595,
      "compression_loss": 0.0,
      "distillation_loss": 0.3017335832118988,
      "epoch": 0.69,
      "learning_rate": 4.883530183102123e-05,
      "loss": 0.29,
      "step": 722,
      "task_loss": 0.1840285062789917
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.723571476734366,
      "compression_loss": 0.0,
      "distillation_loss": 0.27277618646621704,
      "epoch": 0.69,
      "learning_rate": 4.883208528711715e-05,
      "loss": 0.2582,
      "step": 723,
      "task_loss": 0.1269429624080658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.723601242483713,
      "compression_loss": 0.0,
      "distillation_loss": 0.2977296710014343,
      "epoch": 0.69,
      "learning_rate": 4.8828864413999995e-05,
      "loss": 0.2751,
      "step": 724,
      "task_loss": 0.07180985063314438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7236310005037043,
      "compression_loss": 0.0,
      "distillation_loss": 0.08697059750556946,
      "epoch": 0.69,
      "learning_rate": 4.8825639212254865e-05,
      "loss": 0.0934,
      "step": 725,
      "task_loss": 0.15161065757274628
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7236607507953434,
      "compression_loss": 0.0,
      "distillation_loss": 0.190194234251976,
      "epoch": 0.69,
      "learning_rate": 4.882240968246762e-05,
      "loss": 0.1799,
      "step": 726,
      "task_loss": 0.087301105260849
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7236904933596341,
      "compression_loss": 0.0,
      "distillation_loss": 0.22055265307426453,
      "epoch": 0.69,
      "learning_rate": 4.8819175825224925e-05,
      "loss": 0.2109,
      "step": 727,
      "task_loss": 0.12390641123056412
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7237202281975801,
      "compression_loss": 0.0,
      "distillation_loss": 0.18130475282669067,
      "epoch": 0.69,
      "learning_rate": 4.881593764111424e-05,
      "loss": 0.1789,
      "step": 728,
      "task_loss": 0.15703324973583221
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7237499553101849,
      "compression_loss": 0.0,
      "distillation_loss": 0.21481934189796448,
      "epoch": 0.69,
      "learning_rate": 4.8812695130723775e-05,
      "loss": 0.2068,
      "step": 729,
      "task_loss": 0.13483870029449463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7237796746984524,
      "compression_loss": 0.0,
      "distillation_loss": 0.2288239449262619,
      "epoch": 0.69,
      "learning_rate": 4.880944829464256e-05,
      "loss": 0.2152,
      "step": 730,
      "task_loss": 0.09266623109579086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7238093863633862,
      "compression_loss": 0.0,
      "distillation_loss": 0.43135643005371094,
      "epoch": 0.69,
      "learning_rate": 4.880619713346039e-05,
      "loss": 0.4258,
      "step": 731,
      "task_loss": 0.37580642104148865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7238390903059899,
      "compression_loss": 0.0,
      "distillation_loss": 0.5182836055755615,
      "epoch": 0.7,
      "learning_rate": 4.8802941647767856e-05,
      "loss": 0.4974,
      "step": 732,
      "task_loss": 0.3091394901275635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7238687865272674,
      "compression_loss": 0.0,
      "distillation_loss": 0.33886805176734924,
      "epoch": 0.7,
      "learning_rate": 4.879968183815634e-05,
      "loss": 0.3219,
      "step": 733,
      "task_loss": 0.16887077689170837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7238984750282221,
      "compression_loss": 0.0,
      "distillation_loss": 0.16470275819301605,
      "epoch": 0.7,
      "learning_rate": 4.8796417705217994e-05,
      "loss": 0.1544,
      "step": 734,
      "task_loss": 0.06122714653611183
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7239281558098579,
      "compression_loss": 0.0,
      "distillation_loss": 0.3576793074607849,
      "epoch": 0.7,
      "learning_rate": 4.879314924954577e-05,
      "loss": 0.3392,
      "step": 735,
      "task_loss": 0.1729682832956314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7239578288731784,
      "compression_loss": 0.0,
      "distillation_loss": 0.1777103692293167,
      "epoch": 0.7,
      "learning_rate": 4.87898764717334e-05,
      "loss": 0.1658,
      "step": 736,
      "task_loss": 0.0585070438683033
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7239874942191873,
      "compression_loss": 0.0,
      "distillation_loss": 0.21319881081581116,
      "epoch": 0.7,
      "learning_rate": 4.8786599372375384e-05,
      "loss": 0.2003,
      "step": 737,
      "task_loss": 0.08430507779121399
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7240171518488882,
      "compression_loss": 0.0,
      "distillation_loss": 0.4297768473625183,
      "epoch": 0.7,
      "learning_rate": 4.878331795206705e-05,
      "loss": 0.4102,
      "step": 738,
      "task_loss": 0.23426848649978638
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7240468017632848,
      "compression_loss": 0.0,
      "distillation_loss": 0.44237983226776123,
      "epoch": 0.7,
      "learning_rate": 4.878003221140446e-05,
      "loss": 0.4247,
      "step": 739,
      "task_loss": 0.2660304307937622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.724076443963381,
      "compression_loss": 0.0,
      "distillation_loss": 0.47253918647766113,
      "epoch": 0.7,
      "learning_rate": 4.877674215098449e-05,
      "loss": 0.4542,
      "step": 740,
      "task_loss": 0.28894340991973877
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7241060784501803,
      "compression_loss": 0.0,
      "distillation_loss": 0.3578449487686157,
      "epoch": 0.7,
      "learning_rate": 4.87734477714048e-05,
      "loss": 0.3449,
      "step": 741,
      "task_loss": 0.22871339321136475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7241357052246863,
      "compression_loss": 0.0,
      "distillation_loss": 0.41139233112335205,
      "epoch": 0.7,
      "learning_rate": 4.8770149073263833e-05,
      "loss": 0.3942,
      "step": 742,
      "task_loss": 0.23949165642261505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7241653242879028,
      "compression_loss": 0.0,
      "distillation_loss": 0.24485422670841217,
      "epoch": 0.71,
      "learning_rate": 4.87668460571608e-05,
      "loss": 0.2349,
      "step": 743,
      "task_loss": 0.1458071619272232
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7241949356408335,
      "compression_loss": 0.0,
      "distillation_loss": 0.2777606248855591,
      "epoch": 0.71,
      "learning_rate": 4.8763538723695726e-05,
      "loss": 0.2744,
      "step": 744,
      "task_loss": 0.2439776211977005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7242245392844819,
      "compression_loss": 0.0,
      "distillation_loss": 0.4678876996040344,
      "epoch": 0.71,
      "learning_rate": 4.87602270734694e-05,
      "loss": 0.444,
      "step": 745,
      "task_loss": 0.22880345582962036
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7242541352198519,
      "compression_loss": 0.0,
      "distillation_loss": 0.5279219150543213,
      "epoch": 0.71,
      "learning_rate": 4.8756911107083387e-05,
      "loss": 0.5044,
      "step": 746,
      "task_loss": 0.2927531599998474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7242837234479472,
      "compression_loss": 0.0,
      "distillation_loss": 0.35156458616256714,
      "epoch": 0.71,
      "learning_rate": 4.875359082514006e-05,
      "loss": 0.3392,
      "step": 747,
      "task_loss": 0.2275211364030838
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7243133039697713,
      "compression_loss": 0.0,
      "distillation_loss": 0.18168729543685913,
      "epoch": 0.71,
      "learning_rate": 4.8750266228242555e-05,
      "loss": 0.1755,
      "step": 748,
      "task_loss": 0.11939448118209839
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7243428767863281,
      "compression_loss": 0.0,
      "distillation_loss": 0.3357431888580322,
      "epoch": 0.71,
      "learning_rate": 4.874693731699481e-05,
      "loss": 0.3205,
      "step": 749,
      "task_loss": 0.18339495360851288
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.724372441898621,
      "compression_loss": 0.0,
      "distillation_loss": 0.2332255244255066,
      "epoch": 0.71,
      "learning_rate": 4.8743604092001544e-05,
      "loss": 0.2189,
      "step": 750,
      "task_loss": 0.08988796174526215
    },
    {
      "epoch": 0.71,
      "eval_accuracy": 0.8990825688073395,
      "eval_loss": 0.3525860905647278,
      "eval_runtime": 18.2739,
      "eval_samples_per_second": 47.718,
      "eval_steps_per_second": 5.965,
      "step": 750
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.724401999307654,
      "compression_loss": 0.0,
      "distillation_loss": 0.23790621757507324,
      "epoch": 0.71,
      "learning_rate": 4.8740266553868236e-05,
      "loss": 0.2297,
      "step": 751,
      "task_loss": 0.15616914629936218
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7244315490144305,
      "compression_loss": 0.0,
      "distillation_loss": 0.1910121589899063,
      "epoch": 0.71,
      "learning_rate": 4.873692470320117e-05,
      "loss": 0.1767,
      "step": 752,
      "task_loss": 0.04741794615983963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7244610910199543,
      "compression_loss": 0.0,
      "distillation_loss": 0.6017439961433411,
      "epoch": 0.72,
      "learning_rate": 4.8733578540607425e-05,
      "loss": 0.587,
      "step": 753,
      "task_loss": 0.45440027117729187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7244906253252291,
      "compression_loss": 0.0,
      "distillation_loss": 0.1898229718208313,
      "epoch": 0.72,
      "learning_rate": 4.8730228066694825e-05,
      "loss": 0.1778,
      "step": 754,
      "task_loss": 0.06922735273838043
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7245201519312585,
      "compression_loss": 0.0,
      "distillation_loss": 0.1818356066942215,
      "epoch": 0.72,
      "learning_rate": 4.872687328207202e-05,
      "loss": 0.1712,
      "step": 755,
      "task_loss": 0.07538501918315887
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7245496708390464,
      "compression_loss": 0.0,
      "distillation_loss": 0.3183574080467224,
      "epoch": 0.72,
      "learning_rate": 4.872351418734841e-05,
      "loss": 0.3017,
      "step": 756,
      "task_loss": 0.15180720388889313
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7245791820495963,
      "compression_loss": 0.0,
      "distillation_loss": 0.35951346158981323,
      "epoch": 0.72,
      "learning_rate": 4.8720150783134196e-05,
      "loss": 0.3478,
      "step": 757,
      "task_loss": 0.2423841804265976
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7246086855639118,
      "compression_loss": 0.0,
      "distillation_loss": 0.18601301312446594,
      "epoch": 0.72,
      "learning_rate": 4.871678307004035e-05,
      "loss": 0.1736,
      "step": 758,
      "task_loss": 0.06224619597196579
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7246381813829967,
      "compression_loss": 0.0,
      "distillation_loss": 0.3574899733066559,
      "epoch": 0.72,
      "learning_rate": 4.8713411048678635e-05,
      "loss": 0.3498,
      "step": 759,
      "task_loss": 0.280869722366333
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7246676695078548,
      "compression_loss": 0.0,
      "distillation_loss": 0.4299178719520569,
      "epoch": 0.72,
      "learning_rate": 4.8710034719661614e-05,
      "loss": 0.4142,
      "step": 760,
      "task_loss": 0.2722673714160919
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7246971499394895,
      "compression_loss": 0.0,
      "distillation_loss": 0.2085331380367279,
      "epoch": 0.72,
      "learning_rate": 4.870665408360258e-05,
      "loss": 0.1979,
      "step": 761,
      "task_loss": 0.10241978615522385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7247266226789048,
      "compression_loss": 0.0,
      "distillation_loss": 0.44271695613861084,
      "epoch": 0.72,
      "learning_rate": 4.870326914111567e-05,
      "loss": 0.424,
      "step": 762,
      "task_loss": 0.2551548182964325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7247560877271041,
      "compression_loss": 0.0,
      "distillation_loss": 0.36846601963043213,
      "epoch": 0.72,
      "learning_rate": 4.8699879892815756e-05,
      "loss": 0.3636,
      "step": 763,
      "task_loss": 0.31974709033966064
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7247855450850912,
      "compression_loss": 0.0,
      "distillation_loss": 0.3941296935081482,
      "epoch": 0.73,
      "learning_rate": 4.8696486339318524e-05,
      "loss": 0.3761,
      "step": 764,
      "task_loss": 0.21385695040225983
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7248149947538698,
      "compression_loss": 0.0,
      "distillation_loss": 0.26239490509033203,
      "epoch": 0.73,
      "learning_rate": 4.8693088481240424e-05,
      "loss": 0.2546,
      "step": 765,
      "task_loss": 0.18409638106822968
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7248444367344437,
      "compression_loss": 0.0,
      "distillation_loss": 0.1657324880361557,
      "epoch": 0.73,
      "learning_rate": 4.86896863191987e-05,
      "loss": 0.156,
      "step": 766,
      "task_loss": 0.06812982261180878
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7248738710278163,
      "compression_loss": 0.0,
      "distillation_loss": 0.22085291147232056,
      "epoch": 0.73,
      "learning_rate": 4.8686279853811356e-05,
      "loss": 0.2073,
      "step": 767,
      "task_loss": 0.08557181805372238
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7249032976349914,
      "compression_loss": 0.0,
      "distillation_loss": 0.06682948768138885,
      "epoch": 0.73,
      "learning_rate": 4.8682869085697206e-05,
      "loss": 0.0704,
      "step": 768,
      "task_loss": 0.10302991420030594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7249327165569729,
      "compression_loss": 0.0,
      "distillation_loss": 0.22944039106369019,
      "epoch": 0.73,
      "learning_rate": 4.8679454015475835e-05,
      "loss": 0.2132,
      "step": 769,
      "task_loss": 0.06714411824941635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7249621277947641,
      "compression_loss": 0.0,
      "distillation_loss": 0.5438629388809204,
      "epoch": 0.73,
      "learning_rate": 4.867603464376759e-05,
      "loss": 0.5187,
      "step": 770,
      "task_loss": 0.291761577129364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.724991531349369,
      "compression_loss": 0.0,
      "distillation_loss": 0.13500595092773438,
      "epoch": 0.73,
      "learning_rate": 4.867261097119363e-05,
      "loss": 0.1268,
      "step": 771,
      "task_loss": 0.0527961365878582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7250209272217911,
      "compression_loss": 0.0,
      "distillation_loss": 0.2532868981361389,
      "epoch": 0.73,
      "learning_rate": 4.8669182998375884e-05,
      "loss": 0.2471,
      "step": 772,
      "task_loss": 0.19150203466415405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7250503154130341,
      "compression_loss": 0.0,
      "distillation_loss": 0.1466882824897766,
      "epoch": 0.73,
      "learning_rate": 4.8665750725937045e-05,
      "loss": 0.138,
      "step": 773,
      "task_loss": 0.059596575796604156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7250796959241018,
      "compression_loss": 0.0,
      "distillation_loss": 0.19294121861457825,
      "epoch": 0.74,
      "learning_rate": 4.866231415450062e-05,
      "loss": 0.1905,
      "step": 774,
      "task_loss": 0.16841061413288116
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7251090687559978,
      "compression_loss": 0.0,
      "distillation_loss": 0.21601201593875885,
      "epoch": 0.74,
      "learning_rate": 4.8658873284690866e-05,
      "loss": 0.2109,
      "step": 775,
      "task_loss": 0.16467759013175964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7251384339097258,
      "compression_loss": 0.0,
      "distillation_loss": 0.2825254797935486,
      "epoch": 0.74,
      "learning_rate": 4.865542811713284e-05,
      "loss": 0.283,
      "step": 776,
      "task_loss": 0.28761669993400574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7251677913862895,
      "compression_loss": 0.0,
      "distillation_loss": 0.17407667636871338,
      "epoch": 0.74,
      "learning_rate": 4.865197865245237e-05,
      "loss": 0.1644,
      "step": 777,
      "task_loss": 0.07734289765357971
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7251971411866924,
      "compression_loss": 0.0,
      "distillation_loss": 0.1487307846546173,
      "epoch": 0.74,
      "learning_rate": 4.8648524891276066e-05,
      "loss": 0.1402,
      "step": 778,
      "task_loss": 0.06377163529396057
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7252264833119385,
      "compression_loss": 0.0,
      "distillation_loss": 0.6334168910980225,
      "epoch": 0.74,
      "learning_rate": 4.8645066834231325e-05,
      "loss": 0.616,
      "step": 779,
      "task_loss": 0.4591747522354126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7252558177630313,
      "compression_loss": 0.0,
      "distillation_loss": 0.3129650950431824,
      "epoch": 0.74,
      "learning_rate": 4.8641604481946314e-05,
      "loss": 0.2944,
      "step": 780,
      "task_loss": 0.12727110087871552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7252851445409744,
      "compression_loss": 0.0,
      "distillation_loss": 0.5313491821289062,
      "epoch": 0.74,
      "learning_rate": 4.863813783504999e-05,
      "loss": 0.5238,
      "step": 781,
      "task_loss": 0.4554893374443054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7253144636467717,
      "compression_loss": 0.0,
      "distillation_loss": 0.16231770813465118,
      "epoch": 0.74,
      "learning_rate": 4.863466689417209e-05,
      "loss": 0.1523,
      "step": 782,
      "task_loss": 0.062042634934186935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7253437750814267,
      "compression_loss": 0.0,
      "distillation_loss": 0.20974399149417877,
      "epoch": 0.74,
      "learning_rate": 4.863119165994312e-05,
      "loss": 0.2063,
      "step": 783,
      "task_loss": 0.1748623102903366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7253730788459432,
      "compression_loss": 0.0,
      "distillation_loss": 0.2711687684059143,
      "epoch": 0.74,
      "learning_rate": 4.862771213299438e-05,
      "loss": 0.2706,
      "step": 784,
      "task_loss": 0.26576724648475647
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7254023749413246,
      "compression_loss": 0.0,
      "distillation_loss": 0.2579856514930725,
      "epoch": 0.75,
      "learning_rate": 4.8624228313957937e-05,
      "loss": 0.2482,
      "step": 785,
      "task_loss": 0.1598002314567566
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.725431663368575,
      "compression_loss": 0.0,
      "distillation_loss": 0.2370847761631012,
      "epoch": 0.75,
      "learning_rate": 4.862074020346664e-05,
      "loss": 0.2268,
      "step": 786,
      "task_loss": 0.1346910297870636
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7254609441286979,
      "compression_loss": 0.0,
      "distillation_loss": 0.4882653057575226,
      "epoch": 0.75,
      "learning_rate": 4.8617247802154134e-05,
      "loss": 0.4718,
      "step": 787,
      "task_loss": 0.323346883058548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7254902172226969,
      "compression_loss": 0.0,
      "distillation_loss": 0.5949300527572632,
      "epoch": 0.75,
      "learning_rate": 4.861375111065482e-05,
      "loss": 0.5811,
      "step": 788,
      "task_loss": 0.4562370777130127
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7255194826515758,
      "compression_loss": 0.0,
      "distillation_loss": 0.5340683460235596,
      "epoch": 0.75,
      "learning_rate": 4.861025012960389e-05,
      "loss": 0.5078,
      "step": 789,
      "task_loss": 0.27188482880592346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7255487404163383,
      "compression_loss": 0.0,
      "distillation_loss": 0.23920854926109314,
      "epoch": 0.75,
      "learning_rate": 4.8606744859637316e-05,
      "loss": 0.2273,
      "step": 790,
      "task_loss": 0.11962777376174927
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.725577990517988,
      "compression_loss": 0.0,
      "distillation_loss": 0.25565600395202637,
      "epoch": 0.75,
      "learning_rate": 4.8603235301391844e-05,
      "loss": 0.2569,
      "step": 791,
      "task_loss": 0.26804494857788086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7256072329575285,
      "compression_loss": 0.0,
      "distillation_loss": 0.2217859923839569,
      "epoch": 0.75,
      "learning_rate": 4.859972145550501e-05,
      "loss": 0.2133,
      "step": 792,
      "task_loss": 0.13717922568321228
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7256364677359637,
      "compression_loss": 0.0,
      "distillation_loss": 0.24465468525886536,
      "epoch": 0.75,
      "learning_rate": 4.859620332261512e-05,
      "loss": 0.2358,
      "step": 793,
      "task_loss": 0.15646812319755554
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7256656948542971,
      "compression_loss": 0.0,
      "distillation_loss": 0.23842063546180725,
      "epoch": 0.75,
      "learning_rate": 4.8592680903361247e-05,
      "loss": 0.2273,
      "step": 794,
      "task_loss": 0.12698203325271606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7256949143135324,
      "compression_loss": 0.0,
      "distillation_loss": 0.26218464970588684,
      "epoch": 0.75,
      "learning_rate": 4.858915419838327e-05,
      "loss": 0.2558,
      "step": 795,
      "task_loss": 0.19839191436767578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7257241261146734,
      "compression_loss": 0.0,
      "distillation_loss": 0.5854094624519348,
      "epoch": 0.76,
      "learning_rate": 4.8585623208321825e-05,
      "loss": 0.5535,
      "step": 796,
      "task_loss": 0.2661326229572296
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7257533302587237,
      "compression_loss": 0.0,
      "distillation_loss": 0.18930557370185852,
      "epoch": 0.76,
      "learning_rate": 4.858208793381833e-05,
      "loss": 0.1816,
      "step": 797,
      "task_loss": 0.11239316314458847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.725782526746687,
      "compression_loss": 0.0,
      "distillation_loss": 0.35272884368896484,
      "epoch": 0.76,
      "learning_rate": 4.8578548375514995e-05,
      "loss": 0.3343,
      "step": 798,
      "task_loss": 0.16825662553310394
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7258117155795669,
      "compression_loss": 0.0,
      "distillation_loss": 0.6028088331222534,
      "epoch": 0.76,
      "learning_rate": 4.8575004534054794e-05,
      "loss": 0.5743,
      "step": 799,
      "task_loss": 0.31754568219184875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7258408967583673,
      "compression_loss": 0.0,
      "distillation_loss": 0.16018742322921753,
      "epoch": 0.76,
      "learning_rate": 4.8571456410081474e-05,
      "loss": 0.1666,
      "step": 800,
      "task_loss": 0.22434723377227783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7258700702840917,
      "compression_loss": 0.0,
      "distillation_loss": 0.346615195274353,
      "epoch": 0.76,
      "learning_rate": 4.856790400423958e-05,
      "loss": 0.3398,
      "step": 801,
      "task_loss": 0.2779731750488281
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7258992361577438,
      "compression_loss": 0.0,
      "distillation_loss": 0.40023940801620483,
      "epoch": 0.76,
      "learning_rate": 4.856434731717442e-05,
      "loss": 0.388,
      "step": 802,
      "task_loss": 0.27803096175193787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7259283943803273,
      "compression_loss": 0.0,
      "distillation_loss": 0.213174507021904,
      "epoch": 0.76,
      "learning_rate": 4.8560786349532075e-05,
      "loss": 0.199,
      "step": 803,
      "task_loss": 0.07103019952774048
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7259575449528458,
      "compression_loss": 0.0,
      "distillation_loss": 0.15435612201690674,
      "epoch": 0.76,
      "learning_rate": 4.855722110195943e-05,
      "loss": 0.145,
      "step": 804,
      "task_loss": 0.06065506860613823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7259866878763032,
      "compression_loss": 0.0,
      "distillation_loss": 0.20377478003501892,
      "epoch": 0.76,
      "learning_rate": 4.8553651575104114e-05,
      "loss": 0.1985,
      "step": 805,
      "task_loss": 0.15101395547389984
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.726015823151703,
      "compression_loss": 0.0,
      "distillation_loss": 0.3351978063583374,
      "epoch": 0.77,
      "learning_rate": 4.8550077769614554e-05,
      "loss": 0.3193,
      "step": 806,
      "task_loss": 0.17666833102703094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7260449507800489,
      "compression_loss": 0.0,
      "distillation_loss": 0.2128250151872635,
      "epoch": 0.77,
      "learning_rate": 4.8546499686139944e-05,
      "loss": 0.1992,
      "step": 807,
      "task_loss": 0.07669594138860703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7260740707623446,
      "compression_loss": 0.0,
      "distillation_loss": 0.22992324829101562,
      "epoch": 0.77,
      "learning_rate": 4.854291732533027e-05,
      "loss": 0.2204,
      "step": 808,
      "task_loss": 0.13490459322929382
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.726103183099594,
      "compression_loss": 0.0,
      "distillation_loss": 0.1448119878768921,
      "epoch": 0.77,
      "learning_rate": 4.853933068783628e-05,
      "loss": 0.1417,
      "step": 809,
      "task_loss": 0.11348484456539154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7261322877928004,
      "compression_loss": 0.0,
      "distillation_loss": 0.16505393385887146,
      "epoch": 0.77,
      "learning_rate": 4.853573977430951e-05,
      "loss": 0.1539,
      "step": 810,
      "task_loss": 0.053132861852645874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7261613848429677,
      "compression_loss": 0.0,
      "distillation_loss": 0.15083172917366028,
      "epoch": 0.77,
      "learning_rate": 4.8532144585402254e-05,
      "loss": 0.1491,
      "step": 811,
      "task_loss": 0.13310600817203522
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7261904742510995,
      "compression_loss": 0.0,
      "distillation_loss": 0.29990440607070923,
      "epoch": 0.77,
      "learning_rate": 4.85285451217676e-05,
      "loss": 0.2841,
      "step": 812,
      "task_loss": 0.1421201378107071
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7262195560181995,
      "compression_loss": 0.0,
      "distillation_loss": 0.20483794808387756,
      "epoch": 0.77,
      "learning_rate": 4.8524941384059415e-05,
      "loss": 0.1949,
      "step": 813,
      "task_loss": 0.10570985078811646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7262486301452715,
      "compression_loss": 0.0,
      "distillation_loss": 0.152552530169487,
      "epoch": 0.77,
      "learning_rate": 4.8521333372932326e-05,
      "loss": 0.1494,
      "step": 814,
      "task_loss": 0.12140922248363495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7262776966333191,
      "compression_loss": 0.0,
      "distillation_loss": 0.2609613537788391,
      "epoch": 0.77,
      "learning_rate": 4.851772108904175e-05,
      "loss": 0.2615,
      "step": 815,
      "task_loss": 0.2658497095108032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7263067554833459,
      "compression_loss": 0.0,
      "distillation_loss": 0.07736967504024506,
      "epoch": 0.77,
      "learning_rate": 4.851410453304388e-05,
      "loss": 0.0738,
      "step": 816,
      "task_loss": 0.041510533541440964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7263358066963557,
      "compression_loss": 0.0,
      "distillation_loss": 0.17229236662387848,
      "epoch": 0.78,
      "learning_rate": 4.851048370559567e-05,
      "loss": 0.1659,
      "step": 817,
      "task_loss": 0.10811686515808105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7263648502733522,
      "compression_loss": 0.0,
      "distillation_loss": 0.15134498476982117,
      "epoch": 0.78,
      "learning_rate": 4.850685860735487e-05,
      "loss": 0.1597,
      "step": 818,
      "task_loss": 0.23519474267959595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7263938862153388,
      "compression_loss": 0.0,
      "distillation_loss": 0.42961549758911133,
      "epoch": 0.78,
      "learning_rate": 4.850322923898e-05,
      "loss": 0.408,
      "step": 819,
      "task_loss": 0.21310321986675262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7264229145233196,
      "compression_loss": 0.0,
      "distillation_loss": 0.15049859881401062,
      "epoch": 0.78,
      "learning_rate": 4.8499595601130337e-05,
      "loss": 0.1421,
      "step": 820,
      "task_loss": 0.06645572930574417
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.726451935198298,
      "compression_loss": 0.0,
      "distillation_loss": 0.1578487902879715,
      "epoch": 0.78,
      "learning_rate": 4.849595769446596e-05,
      "loss": 0.1482,
      "step": 821,
      "task_loss": 0.06112413853406906
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7264809482412778,
      "compression_loss": 0.0,
      "distillation_loss": 0.26902398467063904,
      "epoch": 0.78,
      "learning_rate": 4.849231551964771e-05,
      "loss": 0.2534,
      "step": 822,
      "task_loss": 0.1128850132226944
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7265099536532627,
      "compression_loss": 0.0,
      "distillation_loss": 0.2599482536315918,
      "epoch": 0.78,
      "learning_rate": 4.848866907733721e-05,
      "loss": 0.2607,
      "step": 823,
      "task_loss": 0.26737767457962036
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7265389514352562,
      "compression_loss": 0.0,
      "distillation_loss": 0.4169207513332367,
      "epoch": 0.78,
      "learning_rate": 4.848501836819684e-05,
      "loss": 0.4028,
      "step": 824,
      "task_loss": 0.2758685052394867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7265679415882622,
      "compression_loss": 0.0,
      "distillation_loss": 0.09598681330680847,
      "epoch": 0.78,
      "learning_rate": 4.848136339288979e-05,
      "loss": 0.0934,
      "step": 825,
      "task_loss": 0.06983894854784012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7265969241132842,
      "compression_loss": 0.0,
      "distillation_loss": 0.7514071464538574,
      "epoch": 0.78,
      "learning_rate": 4.8477704152079984e-05,
      "loss": 0.7297,
      "step": 826,
      "task_loss": 0.5340147614479065
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7266258990113261,
      "compression_loss": 0.0,
      "distillation_loss": 0.41649430990219116,
      "epoch": 0.79,
      "learning_rate": 4.8474040646432153e-05,
      "loss": 0.4065,
      "step": 827,
      "task_loss": 0.31665799021720886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7266548662833914,
      "compression_loss": 0.0,
      "distillation_loss": 0.2038126438856125,
      "epoch": 0.79,
      "learning_rate": 4.8470372876611784e-05,
      "loss": 0.1918,
      "step": 828,
      "task_loss": 0.08404532074928284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7266838259304838,
      "compression_loss": 0.0,
      "distillation_loss": 0.09588038921356201,
      "epoch": 0.79,
      "learning_rate": 4.846670084328515e-05,
      "loss": 0.109,
      "step": 829,
      "task_loss": 0.22751304507255554
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7267127779536071,
      "compression_loss": 0.0,
      "distillation_loss": 0.44214335083961487,
      "epoch": 0.79,
      "learning_rate": 4.846302454711929e-05,
      "loss": 0.4217,
      "step": 830,
      "task_loss": 0.23792800307273865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7267417223537649,
      "compression_loss": 0.0,
      "distillation_loss": 0.23889368772506714,
      "epoch": 0.79,
      "learning_rate": 4.845934398878202e-05,
      "loss": 0.2285,
      "step": 831,
      "task_loss": 0.13461318612098694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7267706591319608,
      "compression_loss": 0.0,
      "distillation_loss": 0.23278114199638367,
      "epoch": 0.79,
      "learning_rate": 4.845565916894193e-05,
      "loss": 0.2227,
      "step": 832,
      "task_loss": 0.13188397884368896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7267995882891986,
      "compression_loss": 0.0,
      "distillation_loss": 0.3519028425216675,
      "epoch": 0.79,
      "learning_rate": 4.8451970088268396e-05,
      "loss": 0.344,
      "step": 833,
      "task_loss": 0.27317169308662415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.726828509826482,
      "compression_loss": 0.0,
      "distillation_loss": 0.19134777784347534,
      "epoch": 0.79,
      "learning_rate": 4.8448276747431545e-05,
      "loss": 0.1843,
      "step": 834,
      "task_loss": 0.12132743000984192
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7268574237448145,
      "compression_loss": 0.0,
      "distillation_loss": 0.5469926595687866,
      "epoch": 0.79,
      "learning_rate": 4.84445791471023e-05,
      "loss": 0.5193,
      "step": 835,
      "task_loss": 0.27034249901771545
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7268863300452001,
      "compression_loss": 0.0,
      "distillation_loss": 0.22652700543403625,
      "epoch": 0.79,
      "learning_rate": 4.8440877287952336e-05,
      "loss": 0.2205,
      "step": 836,
      "task_loss": 0.16659414768218994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7269152287286422,
      "compression_loss": 0.0,
      "distillation_loss": 0.1271471381187439,
      "epoch": 0.79,
      "learning_rate": 4.8437171170654125e-05,
      "loss": 0.1293,
      "step": 837,
      "task_loss": 0.1484624445438385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7269441197961446,
      "compression_loss": 0.0,
      "distillation_loss": 0.2708894908428192,
      "epoch": 0.8,
      "learning_rate": 4.843346079588089e-05,
      "loss": 0.257,
      "step": 838,
      "task_loss": 0.132267564535141
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7269730032487108,
      "compression_loss": 0.0,
      "distillation_loss": 0.3023916482925415,
      "epoch": 0.8,
      "learning_rate": 4.842974616430665e-05,
      "loss": 0.2919,
      "step": 839,
      "task_loss": 0.197329580783844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7270018790873449,
      "compression_loss": 0.0,
      "distillation_loss": 0.30287623405456543,
      "epoch": 0.8,
      "learning_rate": 4.842602727660618e-05,
      "loss": 0.3182,
      "step": 840,
      "task_loss": 0.4560437798500061
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7270307473130502,
      "compression_loss": 0.0,
      "distillation_loss": 0.23289792239665985,
      "epoch": 0.8,
      "learning_rate": 4.842230413345503e-05,
      "loss": 0.2203,
      "step": 841,
      "task_loss": 0.10670986026525497
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7270596079268306,
      "compression_loss": 0.0,
      "distillation_loss": 0.1355813890695572,
      "epoch": 0.8,
      "learning_rate": 4.8418576735529535e-05,
      "loss": 0.1275,
      "step": 842,
      "task_loss": 0.054461102932691574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7270884609296895,
      "compression_loss": 0.0,
      "distillation_loss": 0.20015108585357666,
      "epoch": 0.8,
      "learning_rate": 4.841484508350679e-05,
      "loss": 0.194,
      "step": 843,
      "task_loss": 0.13830646872520447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7271173063226308,
      "compression_loss": 0.0,
      "distillation_loss": 0.2658821940422058,
      "epoch": 0.8,
      "learning_rate": 4.841110917806467e-05,
      "loss": 0.2551,
      "step": 844,
      "task_loss": 0.15778332948684692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7271461441066583,
      "compression_loss": 0.0,
      "distillation_loss": 0.35619670152664185,
      "epoch": 0.8,
      "learning_rate": 4.840736901988182e-05,
      "loss": 0.3489,
      "step": 845,
      "task_loss": 0.2832150459289551
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7271749742827754,
      "compression_loss": 0.0,
      "distillation_loss": 0.35258805751800537,
      "epoch": 0.8,
      "learning_rate": 4.840362460963765e-05,
      "loss": 0.3371,
      "step": 846,
      "task_loss": 0.19759327173233032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7272037968519861,
      "compression_loss": 0.0,
      "distillation_loss": 0.27292773127555847,
      "epoch": 0.8,
      "learning_rate": 4.8399875948012355e-05,
      "loss": 0.2561,
      "step": 847,
      "task_loss": 0.10433943569660187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7272326118152938,
      "compression_loss": 0.0,
      "distillation_loss": 0.44316327571868896,
      "epoch": 0.81,
      "learning_rate": 4.8396123035686906e-05,
      "loss": 0.4312,
      "step": 848,
      "task_loss": 0.323985755443573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7272614191737022,
      "compression_loss": 0.0,
      "distillation_loss": 0.3451952636241913,
      "epoch": 0.81,
      "learning_rate": 4.839236587334303e-05,
      "loss": 0.3296,
      "step": 849,
      "task_loss": 0.1894971877336502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7272902189282151,
      "compression_loss": 0.0,
      "distillation_loss": 0.3476291000843048,
      "epoch": 0.81,
      "learning_rate": 4.8388604461663236e-05,
      "loss": 0.3311,
      "step": 850,
      "task_loss": 0.18212732672691345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7273190110798362,
      "compression_loss": 0.0,
      "distillation_loss": 0.19594484567642212,
      "epoch": 0.81,
      "learning_rate": 4.838483880133079e-05,
      "loss": 0.1966,
      "step": 851,
      "task_loss": 0.20298801362514496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7273477956295691,
      "compression_loss": 0.0,
      "distillation_loss": 0.2919715940952301,
      "epoch": 0.81,
      "learning_rate": 4.8381068893029766e-05,
      "loss": 0.2813,
      "step": 852,
      "task_loss": 0.18507151305675507
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7273765725784175,
      "compression_loss": 0.0,
      "distillation_loss": 0.2512668967247009,
      "epoch": 0.81,
      "learning_rate": 4.837729473744497e-05,
      "loss": 0.2501,
      "step": 853,
      "task_loss": 0.2395791858434677
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7274053419273852,
      "compression_loss": 0.0,
      "distillation_loss": 0.3293880224227905,
      "epoch": 0.81,
      "learning_rate": 4.8373516335261994e-05,
      "loss": 0.3175,
      "step": 854,
      "task_loss": 0.21075962483882904
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7274341036774756,
      "compression_loss": 0.0,
      "distillation_loss": 0.30303874611854553,
      "epoch": 0.81,
      "learning_rate": 4.8369733687167204e-05,
      "loss": 0.2945,
      "step": 855,
      "task_loss": 0.2173597514629364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7274628578296926,
      "compression_loss": 0.0,
      "distillation_loss": 0.1690731942653656,
      "epoch": 0.81,
      "learning_rate": 4.836594679384775e-05,
      "loss": 0.1559,
      "step": 856,
      "task_loss": 0.03761624917387962
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7274916043850399,
      "compression_loss": 0.0,
      "distillation_loss": 0.3940788507461548,
      "epoch": 0.81,
      "learning_rate": 4.836215565599152e-05,
      "loss": 0.383,
      "step": 857,
      "task_loss": 0.2832016944885254
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7275203433445211,
      "compression_loss": 0.0,
      "distillation_loss": 0.49104076623916626,
      "epoch": 0.81,
      "learning_rate": 4.835836027428722e-05,
      "loss": 0.4723,
      "step": 858,
      "task_loss": 0.3033701777458191
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7275490747091399,
      "compression_loss": 0.0,
      "distillation_loss": 0.34213539958000183,
      "epoch": 0.82,
      "learning_rate": 4.8354560649424264e-05,
      "loss": 0.3263,
      "step": 859,
      "task_loss": 0.18411041796207428
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7275777984798999,
      "compression_loss": 0.0,
      "distillation_loss": 0.10528973489999771,
      "epoch": 0.82,
      "learning_rate": 4.8350756782092894e-05,
      "loss": 0.1246,
      "step": 860,
      "task_loss": 0.2985772490501404
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7276065146578049,
      "compression_loss": 0.0,
      "distillation_loss": 0.28886890411376953,
      "epoch": 0.82,
      "learning_rate": 4.8346948672984096e-05,
      "loss": 0.2806,
      "step": 861,
      "task_loss": 0.20596402883529663
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7276352232438587,
      "compression_loss": 0.0,
      "distillation_loss": 0.24927183985710144,
      "epoch": 0.82,
      "learning_rate": 4.8343136322789626e-05,
      "loss": 0.2412,
      "step": 862,
      "task_loss": 0.16882850229740143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7276639242390647,
      "compression_loss": 0.0,
      "distillation_loss": 0.48319730162620544,
      "epoch": 0.82,
      "learning_rate": 4.8339319732202024e-05,
      "loss": 0.4605,
      "step": 863,
      "task_loss": 0.2564105689525604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7276926176444267,
      "compression_loss": 0.0,
      "distillation_loss": 0.272256463766098,
      "epoch": 0.82,
      "learning_rate": 4.83354989019146e-05,
      "loss": 0.261,
      "step": 864,
      "task_loss": 0.16006284952163696
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7277213034609484,
      "compression_loss": 0.0,
      "distillation_loss": 0.43346107006073,
      "epoch": 0.82,
      "learning_rate": 4.83316738326214e-05,
      "loss": 0.4201,
      "step": 865,
      "task_loss": 0.30030304193496704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7277499816896336,
      "compression_loss": 0.0,
      "distillation_loss": 0.2226218283176422,
      "epoch": 0.82,
      "learning_rate": 4.832784452501729e-05,
      "loss": 0.2106,
      "step": 866,
      "task_loss": 0.1027415320277214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7277786523314856,
      "compression_loss": 0.0,
      "distillation_loss": 0.4321751892566681,
      "epoch": 0.82,
      "learning_rate": 4.8324010979797875e-05,
      "loss": 0.4115,
      "step": 867,
      "task_loss": 0.22510364651679993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7278073153875085,
      "compression_loss": 0.0,
      "distillation_loss": 0.4015240967273712,
      "epoch": 0.82,
      "learning_rate": 4.8320173197659534e-05,
      "loss": 0.3924,
      "step": 868,
      "task_loss": 0.310598224401474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7278359708587059,
      "compression_loss": 0.0,
      "distillation_loss": 0.1309269368648529,
      "epoch": 0.83,
      "learning_rate": 4.831633117929942e-05,
      "loss": 0.1323,
      "step": 869,
      "task_loss": 0.1442694216966629
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7278646187460813,
      "compression_loss": 0.0,
      "distillation_loss": 0.34364479780197144,
      "epoch": 0.83,
      "learning_rate": 4.831248492541545e-05,
      "loss": 0.3322,
      "step": 870,
      "task_loss": 0.2290574610233307
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7278932590506385,
      "compression_loss": 0.0,
      "distillation_loss": 0.4297097623348236,
      "epoch": 0.83,
      "learning_rate": 4.830863443670632e-05,
      "loss": 0.4357,
      "step": 871,
      "task_loss": 0.48918217420578003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7279218917733812,
      "compression_loss": 0.0,
      "distillation_loss": 0.3155050277709961,
      "epoch": 0.83,
      "learning_rate": 4.8304779713871495e-05,
      "loss": 0.3033,
      "step": 872,
      "task_loss": 0.19333837926387787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7279505169153131,
      "compression_loss": 0.0,
      "distillation_loss": 0.19998089969158173,
      "epoch": 0.83,
      "learning_rate": 4.83009207576112e-05,
      "loss": 0.1878,
      "step": 873,
      "task_loss": 0.07784054428339005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7279791344774377,
      "compression_loss": 0.0,
      "distillation_loss": 0.35912150144577026,
      "epoch": 0.83,
      "learning_rate": 4.829705756862642e-05,
      "loss": 0.3322,
      "step": 874,
      "task_loss": 0.09007196873426437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7280077444607589,
      "compression_loss": 0.0,
      "distillation_loss": 0.42145901918411255,
      "epoch": 0.83,
      "learning_rate": 4.829319014761894e-05,
      "loss": 0.4086,
      "step": 875,
      "task_loss": 0.29324570298194885
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7280363468662803,
      "compression_loss": 0.0,
      "distillation_loss": 0.5021111369132996,
      "epoch": 0.83,
      "learning_rate": 4.828931849529129e-05,
      "loss": 0.4807,
      "step": 876,
      "task_loss": 0.28812965750694275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7280649416950056,
      "compression_loss": 0.0,
      "distillation_loss": 0.33752530813217163,
      "epoch": 0.83,
      "learning_rate": 4.8285442612346774e-05,
      "loss": 0.318,
      "step": 877,
      "task_loss": 0.14252959191799164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7280935289479383,
      "compression_loss": 0.0,
      "distillation_loss": 0.24429580569267273,
      "epoch": 0.83,
      "learning_rate": 4.828156249948946e-05,
      "loss": 0.2482,
      "step": 878,
      "task_loss": 0.2833808660507202
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7281221086260824,
      "compression_loss": 0.0,
      "distillation_loss": 0.16154003143310547,
      "epoch": 0.83,
      "learning_rate": 4.827767815742419e-05,
      "loss": 0.1566,
      "step": 879,
      "task_loss": 0.11249750107526779
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7281506807304414,
      "compression_loss": 0.0,
      "distillation_loss": 0.2341882586479187,
      "epoch": 0.84,
      "learning_rate": 4.8273789586856574e-05,
      "loss": 0.2249,
      "step": 880,
      "task_loss": 0.14081409573554993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.728179245262019,
      "compression_loss": 0.0,
      "distillation_loss": 0.22025887668132782,
      "epoch": 0.84,
      "learning_rate": 4.8269896788493e-05,
      "loss": 0.2063,
      "step": 881,
      "task_loss": 0.0802309438586235
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7282078022218189,
      "compression_loss": 0.0,
      "distillation_loss": 0.2393018901348114,
      "epoch": 0.84,
      "learning_rate": 4.8265999763040603e-05,
      "loss": 0.2291,
      "step": 882,
      "task_loss": 0.13759064674377441
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7282363516108448,
      "compression_loss": 0.0,
      "distillation_loss": 0.22338096797466278,
      "epoch": 0.84,
      "learning_rate": 4.8262098511207295e-05,
      "loss": 0.2188,
      "step": 883,
      "task_loss": 0.17738208174705505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7282648934301004,
      "compression_loss": 0.0,
      "distillation_loss": 0.250945508480072,
      "epoch": 0.84,
      "learning_rate": 4.825819303370177e-05,
      "loss": 0.2357,
      "step": 884,
      "task_loss": 0.0986076220870018
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7282934276805891,
      "compression_loss": 0.0,
      "distillation_loss": 0.32160013914108276,
      "epoch": 0.84,
      "learning_rate": 4.8254283331233464e-05,
      "loss": 0.3065,
      "step": 885,
      "task_loss": 0.1702209860086441
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.728321954363315,
      "compression_loss": 0.0,
      "distillation_loss": 0.13798242807388306,
      "epoch": 0.84,
      "learning_rate": 4.825036940451259e-05,
      "loss": 0.1329,
      "step": 886,
      "task_loss": 0.08686836808919907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7283504734792816,
      "compression_loss": 0.0,
      "distillation_loss": 0.28318336606025696,
      "epoch": 0.84,
      "learning_rate": 4.8246451254250145e-05,
      "loss": 0.2692,
      "step": 887,
      "task_loss": 0.14343897998332977
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7283789850294926,
      "compression_loss": 0.0,
      "distillation_loss": 0.31278300285339355,
      "epoch": 0.84,
      "learning_rate": 4.8242528881157866e-05,
      "loss": 0.306,
      "step": 888,
      "task_loss": 0.24526852369308472
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7284074890149516,
      "compression_loss": 0.0,
      "distillation_loss": 0.5371044874191284,
      "epoch": 0.84,
      "learning_rate": 4.823860228594829e-05,
      "loss": 0.5197,
      "step": 889,
      "task_loss": 0.36329126358032227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7284359854366623,
      "compression_loss": 0.0,
      "distillation_loss": 0.19384345412254333,
      "epoch": 0.85,
      "learning_rate": 4.823467146933468e-05,
      "loss": 0.1807,
      "step": 890,
      "task_loss": 0.06265727430582047
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7284644742956286,
      "compression_loss": 0.0,
      "distillation_loss": 0.37346696853637695,
      "epoch": 0.85,
      "learning_rate": 4.823073643203111e-05,
      "loss": 0.3685,
      "step": 891,
      "task_loss": 0.3242899179458618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7284929555928539,
      "compression_loss": 0.0,
      "distillation_loss": 0.4346903860569,
      "epoch": 0.85,
      "learning_rate": 4.822679717475237e-05,
      "loss": 0.4304,
      "step": 892,
      "task_loss": 0.39173489809036255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7285214293293419,
      "compression_loss": 0.0,
      "distillation_loss": 0.36973273754119873,
      "epoch": 0.85,
      "learning_rate": 4.8222853698214076e-05,
      "loss": 0.3513,
      "step": 893,
      "task_loss": 0.18498845398426056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7285498955060965,
      "compression_loss": 0.0,
      "distillation_loss": 0.3774262070655823,
      "epoch": 0.85,
      "learning_rate": 4.8218906003132555e-05,
      "loss": 0.3675,
      "step": 894,
      "task_loss": 0.278170645236969
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7285783541241213,
      "compression_loss": 0.0,
      "distillation_loss": 0.43660831451416016,
      "epoch": 0.85,
      "learning_rate": 4.8214954090224946e-05,
      "loss": 0.4236,
      "step": 895,
      "task_loss": 0.3061750829219818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7286068051844199,
      "compression_loss": 0.0,
      "distillation_loss": 0.2611873149871826,
      "epoch": 0.85,
      "learning_rate": 4.8210997960209114e-05,
      "loss": 0.2515,
      "step": 896,
      "task_loss": 0.16423273086547852
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7286352486879959,
      "compression_loss": 0.0,
      "distillation_loss": 0.2831340730190277,
      "epoch": 0.85,
      "learning_rate": 4.8207037613803715e-05,
      "loss": 0.2735,
      "step": 897,
      "task_loss": 0.18677663803100586
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7286636846358533,
      "compression_loss": 0.0,
      "distillation_loss": 0.38769102096557617,
      "epoch": 0.85,
      "learning_rate": 4.820307305172818e-05,
      "loss": 0.3674,
      "step": 898,
      "task_loss": 0.1845768690109253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7286921130289954,
      "compression_loss": 0.0,
      "distillation_loss": 0.28008437156677246,
      "epoch": 0.85,
      "learning_rate": 4.8199104274702666e-05,
      "loss": 0.265,
      "step": 899,
      "task_loss": 0.1291656196117401
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7287205338684262,
      "compression_loss": 0.0,
      "distillation_loss": 0.12043634802103043,
      "epoch": 0.85,
      "learning_rate": 4.819513128344814e-05,
      "loss": 0.1209,
      "step": 900,
      "task_loss": 0.12554529309272766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7287489471551492,
      "compression_loss": 0.0,
      "distillation_loss": 0.2257360816001892,
      "epoch": 0.86,
      "learning_rate": 4.8191154078686306e-05,
      "loss": 0.2282,
      "step": 901,
      "task_loss": 0.25028055906295776
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7287773528901681,
      "compression_loss": 0.0,
      "distillation_loss": 0.2224644273519516,
      "epoch": 0.86,
      "learning_rate": 4.8187172661139636e-05,
      "loss": 0.2194,
      "step": 902,
      "task_loss": 0.19155338406562805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7288057510744866,
      "compression_loss": 0.0,
      "distillation_loss": 0.1225384920835495,
      "epoch": 0.86,
      "learning_rate": 4.818318703153139e-05,
      "loss": 0.1325,
      "step": 903,
      "task_loss": 0.2224058359861374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7288341417091085,
      "compression_loss": 0.0,
      "distillation_loss": 0.5352866649627686,
      "epoch": 0.86,
      "learning_rate": 4.817919719058557e-05,
      "loss": 0.528,
      "step": 904,
      "task_loss": 0.4623543620109558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7288625247950373,
      "compression_loss": 0.0,
      "distillation_loss": 0.2698153257369995,
      "epoch": 0.86,
      "learning_rate": 4.8175203139026934e-05,
      "loss": 0.2603,
      "step": 905,
      "task_loss": 0.17438843846321106
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7288909003332769,
      "compression_loss": 0.0,
      "distillation_loss": 0.7470159530639648,
      "epoch": 0.86,
      "learning_rate": 4.817120487758104e-05,
      "loss": 0.712,
      "step": 906,
      "task_loss": 0.39666780829429626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7289192683248307,
      "compression_loss": 0.0,
      "distillation_loss": 0.22724510729312897,
      "epoch": 0.86,
      "learning_rate": 4.81672024069742e-05,
      "loss": 0.2208,
      "step": 907,
      "task_loss": 0.16269180178642273
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7289476287707025,
      "compression_loss": 0.0,
      "distillation_loss": 0.3366519808769226,
      "epoch": 0.86,
      "learning_rate": 4.816319572793345e-05,
      "loss": 0.3336,
      "step": 908,
      "task_loss": 0.3060190975666046
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.728975981671896,
      "compression_loss": 0.0,
      "distillation_loss": 0.1925545036792755,
      "epoch": 0.86,
      "learning_rate": 4.815918484118665e-05,
      "loss": 0.18,
      "step": 909,
      "task_loss": 0.0666121393442154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.729004327029415,
      "compression_loss": 0.0,
      "distillation_loss": 0.1643257886171341,
      "epoch": 0.86,
      "learning_rate": 4.815516974746239e-05,
      "loss": 0.151,
      "step": 910,
      "task_loss": 0.030893657356500626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.729032664844263,
      "compression_loss": 0.0,
      "distillation_loss": 0.20738086104393005,
      "epoch": 0.87,
      "learning_rate": 4.815115044749003e-05,
      "loss": 0.1996,
      "step": 911,
      "task_loss": 0.12986215949058533
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7290609951174438,
      "compression_loss": 0.0,
      "distillation_loss": 0.2494029700756073,
      "epoch": 0.87,
      "learning_rate": 4.814712694199969e-05,
      "loss": 0.2375,
      "step": 912,
      "task_loss": 0.13066712021827698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.729089317849961,
      "compression_loss": 0.0,
      "distillation_loss": 0.24174784123897552,
      "epoch": 0.87,
      "learning_rate": 4.814309923172227e-05,
      "loss": 0.2271,
      "step": 913,
      "task_loss": 0.09568721055984497
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7291176330428183,
      "compression_loss": 0.0,
      "distillation_loss": 0.4642338752746582,
      "epoch": 0.87,
      "learning_rate": 4.81390673173894e-05,
      "loss": 0.4392,
      "step": 914,
      "task_loss": 0.2140159010887146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7291459406970194,
      "compression_loss": 0.0,
      "distillation_loss": 0.28287458419799805,
      "epoch": 0.87,
      "learning_rate": 4.8135031199733524e-05,
      "loss": 0.2737,
      "step": 915,
      "task_loss": 0.19124047458171844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7291742408135681,
      "compression_loss": 0.0,
      "distillation_loss": 0.29212427139282227,
      "epoch": 0.87,
      "learning_rate": 4.813099087948781e-05,
      "loss": 0.281,
      "step": 916,
      "task_loss": 0.18083734810352325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7292025333934679,
      "compression_loss": 0.0,
      "distillation_loss": 0.2220340222120285,
      "epoch": 0.87,
      "learning_rate": 4.812694635738621e-05,
      "loss": 0.2106,
      "step": 917,
      "task_loss": 0.10801569372415543
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7292308184377224,
      "compression_loss": 0.0,
      "distillation_loss": 0.1814860701560974,
      "epoch": 0.87,
      "learning_rate": 4.812289763416341e-05,
      "loss": 0.1735,
      "step": 918,
      "task_loss": 0.10150802880525589
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7292590959473356,
      "compression_loss": 0.0,
      "distillation_loss": 0.1173454150557518,
      "epoch": 0.87,
      "learning_rate": 4.81188447105549e-05,
      "loss": 0.11,
      "step": 919,
      "task_loss": 0.0437774695456028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7292873659233109,
      "compression_loss": 0.0,
      "distillation_loss": 0.28012198209762573,
      "epoch": 0.87,
      "learning_rate": 4.811478758729691e-05,
      "loss": 0.2636,
      "step": 920,
      "task_loss": 0.11441508680582047
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7293156283666522,
      "compression_loss": 0.0,
      "distillation_loss": 0.29111599922180176,
      "epoch": 0.87,
      "learning_rate": 4.811072626512642e-05,
      "loss": 0.2741,
      "step": 921,
      "task_loss": 0.12139546126127243
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.729343883278363,
      "compression_loss": 0.0,
      "distillation_loss": 0.36805009841918945,
      "epoch": 0.88,
      "learning_rate": 4.810666074478121e-05,
      "loss": 0.3586,
      "step": 922,
      "task_loss": 0.2733895480632782
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7293721306594471,
      "compression_loss": 0.0,
      "distillation_loss": 0.2743581533432007,
      "epoch": 0.88,
      "learning_rate": 4.8102591026999796e-05,
      "loss": 0.2591,
      "step": 923,
      "task_loss": 0.12169023603200912
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7294003705109081,
      "compression_loss": 0.0,
      "distillation_loss": 0.12222736328840256,
      "epoch": 0.88,
      "learning_rate": 4.8098517112521456e-05,
      "loss": 0.1283,
      "step": 924,
      "task_loss": 0.18253932893276215
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7294286028337497,
      "compression_loss": 0.0,
      "distillation_loss": 0.38871803879737854,
      "epoch": 0.88,
      "learning_rate": 4.8094439002086234e-05,
      "loss": 0.3813,
      "step": 925,
      "task_loss": 0.31461140513420105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7294568276289756,
      "compression_loss": 0.0,
      "distillation_loss": 0.07011576741933823,
      "epoch": 0.88,
      "learning_rate": 4.809035669643495e-05,
      "loss": 0.0841,
      "step": 926,
      "task_loss": 0.21029669046401978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7294850448975896,
      "compression_loss": 0.0,
      "distillation_loss": 0.2029143124818802,
      "epoch": 0.88,
      "learning_rate": 4.808627019630917e-05,
      "loss": 0.1999,
      "step": 927,
      "task_loss": 0.17250923812389374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7295132546405951,
      "compression_loss": 0.0,
      "distillation_loss": 0.4461289942264557,
      "epoch": 0.88,
      "learning_rate": 4.808217950245122e-05,
      "loss": 0.4298,
      "step": 928,
      "task_loss": 0.28285056352615356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7295414568589961,
      "compression_loss": 0.0,
      "distillation_loss": 0.2739975154399872,
      "epoch": 0.88,
      "learning_rate": 4.807808461560419e-05,
      "loss": 0.2818,
      "step": 929,
      "task_loss": 0.35219091176986694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.729569651553796,
      "compression_loss": 0.0,
      "distillation_loss": 0.14522334933280945,
      "epoch": 0.88,
      "learning_rate": 4.8073985536511956e-05,
      "loss": 0.1395,
      "step": 930,
      "task_loss": 0.08773127943277359
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7295978387259987,
      "compression_loss": 0.0,
      "distillation_loss": 0.1583382934331894,
      "epoch": 0.88,
      "learning_rate": 4.806988226591912e-05,
      "loss": 0.1488,
      "step": 931,
      "task_loss": 0.06294950842857361
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7296260183766078,
      "compression_loss": 0.0,
      "distillation_loss": 0.42467373609542847,
      "epoch": 0.89,
      "learning_rate": 4.806577480457106e-05,
      "loss": 0.4083,
      "step": 932,
      "task_loss": 0.26088839769363403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7296541905066269,
      "compression_loss": 0.0,
      "distillation_loss": 0.2598496377468109,
      "epoch": 0.89,
      "learning_rate": 4.8061663153213935e-05,
      "loss": 0.2517,
      "step": 933,
      "task_loss": 0.17828907072544098
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7296823551170598,
      "compression_loss": 0.0,
      "distillation_loss": 0.26050207018852234,
      "epoch": 0.89,
      "learning_rate": 4.805754731259462e-05,
      "loss": 0.2466,
      "step": 934,
      "task_loss": 0.12179625779390335
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7297105122089101,
      "compression_loss": 0.0,
      "distillation_loss": 0.2224908173084259,
      "epoch": 0.89,
      "learning_rate": 4.805342728346079e-05,
      "loss": 0.212,
      "step": 935,
      "task_loss": 0.11794877797365189
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7297386617831816,
      "compression_loss": 0.0,
      "distillation_loss": 0.2159128487110138,
      "epoch": 0.89,
      "learning_rate": 4.804930306656087e-05,
      "loss": 0.2034,
      "step": 936,
      "task_loss": 0.090658038854599
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7297668038408779,
      "compression_loss": 0.0,
      "distillation_loss": 0.42296963930130005,
      "epoch": 0.89,
      "learning_rate": 4.804517466264405e-05,
      "loss": 0.4042,
      "step": 937,
      "task_loss": 0.2355644851922989
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7297949383830027,
      "compression_loss": 0.0,
      "distillation_loss": 0.1279992312192917,
      "epoch": 0.89,
      "learning_rate": 4.8041042072460244e-05,
      "loss": 0.1309,
      "step": 938,
      "task_loss": 0.15678860247135162
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7298230654105596,
      "compression_loss": 0.0,
      "distillation_loss": 0.4838896095752716,
      "epoch": 0.89,
      "learning_rate": 4.803690529676019e-05,
      "loss": 0.4677,
      "step": 939,
      "task_loss": 0.3221690356731415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7298511849245524,
      "compression_loss": 0.0,
      "distillation_loss": 0.3270692229270935,
      "epoch": 0.89,
      "learning_rate": 4.803276433629534e-05,
      "loss": 0.3116,
      "step": 940,
      "task_loss": 0.17188920080661774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7298792969259846,
      "compression_loss": 0.0,
      "distillation_loss": 0.1887792944908142,
      "epoch": 0.89,
      "learning_rate": 4.802861919181793e-05,
      "loss": 0.1795,
      "step": 941,
      "task_loss": 0.09571507573127747
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7299074014158602,
      "compression_loss": 0.0,
      "distillation_loss": 0.33935946226119995,
      "epoch": 0.89,
      "learning_rate": 4.802446986408093e-05,
      "loss": 0.3392,
      "step": 942,
      "task_loss": 0.33791565895080566
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7299354983951827,
      "compression_loss": 0.0,
      "distillation_loss": 0.33818504214286804,
      "epoch": 0.9,
      "learning_rate": 4.8020316353838095e-05,
      "loss": 0.3372,
      "step": 943,
      "task_loss": 0.3283797800540924
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7299635878649556,
      "compression_loss": 0.0,
      "distillation_loss": 0.10457701981067657,
      "epoch": 0.9,
      "learning_rate": 4.8016158661843926e-05,
      "loss": 0.1072,
      "step": 944,
      "task_loss": 0.13106907904148102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7299916698261829,
      "compression_loss": 0.0,
      "distillation_loss": 0.1351221650838852,
      "epoch": 0.9,
      "learning_rate": 4.8011996788853686e-05,
      "loss": 0.1263,
      "step": 945,
      "task_loss": 0.04674810171127319
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7300197442798682,
      "compression_loss": 0.0,
      "distillation_loss": 0.3631080389022827,
      "epoch": 0.9,
      "learning_rate": 4.80078307356234e-05,
      "loss": 0.3498,
      "step": 946,
      "task_loss": 0.2304990440607071
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.730047811227015,
      "compression_loss": 0.0,
      "distillation_loss": 0.42994534969329834,
      "epoch": 0.9,
      "learning_rate": 4.800366050290986e-05,
      "loss": 0.4158,
      "step": 947,
      "task_loss": 0.28830981254577637
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7300758706686271,
      "compression_loss": 0.0,
      "distillation_loss": 0.3227723240852356,
      "epoch": 0.9,
      "learning_rate": 4.799948609147061e-05,
      "loss": 0.3099,
      "step": 948,
      "task_loss": 0.19404058158397675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7301039226057082,
      "compression_loss": 0.0,
      "distillation_loss": 0.20503823459148407,
      "epoch": 0.9,
      "learning_rate": 4.7995307502063936e-05,
      "loss": 0.1954,
      "step": 949,
      "task_loss": 0.10874759405851364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7301319670392621,
      "compression_loss": 0.0,
      "distillation_loss": 0.3428541421890259,
      "epoch": 0.9,
      "learning_rate": 4.799112473544891e-05,
      "loss": 0.3248,
      "step": 950,
      "task_loss": 0.16210930049419403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7301600039702922,
      "compression_loss": 0.0,
      "distillation_loss": 0.31048500537872314,
      "epoch": 0.9,
      "learning_rate": 4.7986937792385344e-05,
      "loss": 0.2987,
      "step": 951,
      "task_loss": 0.19261281192302704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7301880333998025,
      "compression_loss": 0.0,
      "distillation_loss": 0.33201491832733154,
      "epoch": 0.9,
      "learning_rate": 4.798274667363383e-05,
      "loss": 0.3181,
      "step": 952,
      "task_loss": 0.19296838343143463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7302160553287964,
      "compression_loss": 0.0,
      "distillation_loss": 0.4270634949207306,
      "epoch": 0.91,
      "learning_rate": 4.7978551379955684e-05,
      "loss": 0.4057,
      "step": 953,
      "task_loss": 0.21361877024173737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7302440697582777,
      "compression_loss": 0.0,
      "distillation_loss": 0.12916311621665955,
      "epoch": 0.91,
      "learning_rate": 4.797435191211302e-05,
      "loss": 0.1248,
      "step": 954,
      "task_loss": 0.08600229024887085
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7302720766892501,
      "compression_loss": 0.0,
      "distillation_loss": 0.17904284596443176,
      "epoch": 0.91,
      "learning_rate": 4.797014827086869e-05,
      "loss": 0.1746,
      "step": 955,
      "task_loss": 0.13418468832969666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7303000761227173,
      "compression_loss": 0.0,
      "distillation_loss": 0.12082010507583618,
      "epoch": 0.91,
      "learning_rate": 4.79659404569863e-05,
      "loss": 0.1303,
      "step": 956,
      "task_loss": 0.21565213799476624
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7303280680596829,
      "compression_loss": 0.0,
      "distillation_loss": 0.3740869164466858,
      "epoch": 0.91,
      "learning_rate": 4.7961728471230214e-05,
      "loss": 0.3668,
      "step": 957,
      "task_loss": 0.30169612169265747
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7303560525011507,
      "compression_loss": 0.0,
      "distillation_loss": 0.2538146674633026,
      "epoch": 0.91,
      "learning_rate": 4.7957512314365574e-05,
      "loss": 0.2491,
      "step": 958,
      "task_loss": 0.20712712407112122
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7303840294481242,
      "compression_loss": 0.0,
      "distillation_loss": 0.3629278540611267,
      "epoch": 0.91,
      "learning_rate": 4.7953291987158254e-05,
      "loss": 0.3478,
      "step": 959,
      "task_loss": 0.21124333143234253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7304119989016074,
      "compression_loss": 0.0,
      "distillation_loss": 0.2513018846511841,
      "epoch": 0.91,
      "learning_rate": 4.79490674903749e-05,
      "loss": 0.2443,
      "step": 960,
      "task_loss": 0.18155395984649658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7304399608626035,
      "compression_loss": 0.0,
      "distillation_loss": 0.14075066149234772,
      "epoch": 0.91,
      "learning_rate": 4.7944838824782916e-05,
      "loss": 0.1426,
      "step": 961,
      "task_loss": 0.15932926535606384
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7304679153321166,
      "compression_loss": 0.0,
      "distillation_loss": 0.25464701652526855,
      "epoch": 0.91,
      "learning_rate": 4.794060599115045e-05,
      "loss": 0.2359,
      "step": 962,
      "task_loss": 0.06738609075546265
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7304958623111502,
      "compression_loss": 0.0,
      "distillation_loss": 0.12189721316099167,
      "epoch": 0.91,
      "learning_rate": 4.793636899024643e-05,
      "loss": 0.1161,
      "step": 963,
      "task_loss": 0.06351499259471893
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7305238018007081,
      "compression_loss": 0.0,
      "distillation_loss": 0.13325002789497375,
      "epoch": 0.92,
      "learning_rate": 4.7932127822840516e-05,
      "loss": 0.1313,
      "step": 964,
      "task_loss": 0.11407996714115143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7305517338017938,
      "compression_loss": 0.0,
      "distillation_loss": 0.13893647491931915,
      "epoch": 0.92,
      "learning_rate": 4.792788248970314e-05,
      "loss": 0.1442,
      "step": 965,
      "task_loss": 0.1910799741744995
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7305796583154112,
      "compression_loss": 0.0,
      "distillation_loss": 0.49337151646614075,
      "epoch": 0.92,
      "learning_rate": 4.79236329916055e-05,
      "loss": 0.4772,
      "step": 966,
      "task_loss": 0.3313041031360626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7306075753425638,
      "compression_loss": 0.0,
      "distillation_loss": 0.1377095878124237,
      "epoch": 0.92,
      "learning_rate": 4.79193793293195e-05,
      "loss": 0.1368,
      "step": 967,
      "task_loss": 0.12868079543113708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7306354848842553,
      "compression_loss": 0.0,
      "distillation_loss": 0.5649617910385132,
      "epoch": 0.92,
      "learning_rate": 4.791512150361788e-05,
      "loss": 0.5411,
      "step": 968,
      "task_loss": 0.3263978660106659
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7306633869414896,
      "compression_loss": 0.0,
      "distillation_loss": 0.19388450682163239,
      "epoch": 0.92,
      "learning_rate": 4.791085951527408e-05,
      "loss": 0.1838,
      "step": 969,
      "task_loss": 0.09344847500324249
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.73069128151527,
      "compression_loss": 0.0,
      "distillation_loss": 0.11812691390514374,
      "epoch": 0.92,
      "learning_rate": 4.7906593365062304e-05,
      "loss": 0.1124,
      "step": 970,
      "task_loss": 0.06074811518192291
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7307191686066006,
      "compression_loss": 0.0,
      "distillation_loss": 0.2905130386352539,
      "epoch": 0.92,
      "learning_rate": 4.790232305375752e-05,
      "loss": 0.2802,
      "step": 971,
      "task_loss": 0.18719345331192017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7307470482164847,
      "compression_loss": 0.0,
      "distillation_loss": 0.41280168294906616,
      "epoch": 0.92,
      "learning_rate": 4.789804858213547e-05,
      "loss": 0.3943,
      "step": 972,
      "task_loss": 0.2281726449728012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7307749203459263,
      "compression_loss": 0.0,
      "distillation_loss": 0.5764849185943604,
      "epoch": 0.92,
      "learning_rate": 4.7893769950972605e-05,
      "loss": 0.5575,
      "step": 973,
      "task_loss": 0.3868526816368103
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7308027849959289,
      "compression_loss": 0.0,
      "distillation_loss": 0.3850351572036743,
      "epoch": 0.92,
      "learning_rate": 4.788948716104618e-05,
      "loss": 0.3583,
      "step": 974,
      "task_loss": 0.1175268292427063
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7308306421674962,
      "compression_loss": 0.0,
      "distillation_loss": 0.22594526410102844,
      "epoch": 0.93,
      "learning_rate": 4.7885200213134164e-05,
      "loss": 0.2176,
      "step": 975,
      "task_loss": 0.1424587517976761
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7308584918616319,
      "compression_loss": 0.0,
      "distillation_loss": 0.2664077579975128,
      "epoch": 0.93,
      "learning_rate": 4.788090910801532e-05,
      "loss": 0.2586,
      "step": 976,
      "task_loss": 0.18784651160240173
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7308863340793397,
      "compression_loss": 0.0,
      "distillation_loss": 0.24012893438339233,
      "epoch": 0.93,
      "learning_rate": 4.787661384646913e-05,
      "loss": 0.2305,
      "step": 977,
      "task_loss": 0.14341062307357788
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7309141688216232,
      "compression_loss": 0.0,
      "distillation_loss": 0.19192594289779663,
      "epoch": 0.93,
      "learning_rate": 4.787231442927587e-05,
      "loss": 0.1862,
      "step": 978,
      "task_loss": 0.13485443592071533
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7309419960894863,
      "compression_loss": 0.0,
      "distillation_loss": 0.2409208118915558,
      "epoch": 0.93,
      "learning_rate": 4.786801085721654e-05,
      "loss": 0.2312,
      "step": 979,
      "task_loss": 0.14332106709480286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7309698158839324,
      "compression_loss": 0.0,
      "distillation_loss": 0.5161716938018799,
      "epoch": 0.93,
      "learning_rate": 4.78637031310729e-05,
      "loss": 0.4927,
      "step": 980,
      "task_loss": 0.28135231137275696
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7309976282059654,
      "compression_loss": 0.0,
      "distillation_loss": 0.3326793909072876,
      "epoch": 0.93,
      "learning_rate": 4.7859391251627474e-05,
      "loss": 0.3267,
      "step": 981,
      "task_loss": 0.2727680802345276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7310254330565888,
      "compression_loss": 0.0,
      "distillation_loss": 0.1554919183254242,
      "epoch": 0.93,
      "learning_rate": 4.7855075219663535e-05,
      "loss": 0.148,
      "step": 982,
      "task_loss": 0.08066565543413162
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7310532304368065,
      "compression_loss": 0.0,
      "distillation_loss": 0.34702181816101074,
      "epoch": 0.93,
      "learning_rate": 4.785075503596511e-05,
      "loss": 0.3353,
      "step": 983,
      "task_loss": 0.2300397753715515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.731081020347622,
      "compression_loss": 0.0,
      "distillation_loss": 0.07944006472826004,
      "epoch": 0.93,
      "learning_rate": 4.7846430701316994e-05,
      "loss": 0.0734,
      "step": 984,
      "task_loss": 0.018551897257566452
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7311088027900391,
      "compression_loss": 0.0,
      "distillation_loss": 0.35553303360939026,
      "epoch": 0.94,
      "learning_rate": 4.78421022165047e-05,
      "loss": 0.3459,
      "step": 985,
      "task_loss": 0.25915029644966125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7311365777650615,
      "compression_loss": 0.0,
      "distillation_loss": 0.2604832351207733,
      "epoch": 0.94,
      "learning_rate": 4.783776958231453e-05,
      "loss": 0.2522,
      "step": 986,
      "task_loss": 0.1778934746980667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7311643452736927,
      "compression_loss": 0.0,
      "distillation_loss": 0.4339986741542816,
      "epoch": 0.94,
      "learning_rate": 4.783343279953353e-05,
      "loss": 0.4111,
      "step": 987,
      "task_loss": 0.205443874001503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7311921053169365,
      "compression_loss": 0.0,
      "distillation_loss": 0.09336867183446884,
      "epoch": 0.94,
      "learning_rate": 4.782909186894949e-05,
      "loss": 0.0875,
      "step": 988,
      "task_loss": 0.03483529016375542
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7312198578957965,
      "compression_loss": 0.0,
      "distillation_loss": 0.3490631878376007,
      "epoch": 0.94,
      "learning_rate": 4.782474679135097e-05,
      "loss": 0.3363,
      "step": 989,
      "task_loss": 0.22176185250282288
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7312476030112766,
      "compression_loss": 0.0,
      "distillation_loss": 0.31977570056915283,
      "epoch": 0.94,
      "learning_rate": 4.782039756752727e-05,
      "loss": 0.3017,
      "step": 990,
      "task_loss": 0.13884237408638
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7312753406643802,
      "compression_loss": 0.0,
      "distillation_loss": 0.2667671740055084,
      "epoch": 0.94,
      "learning_rate": 4.781604419826845e-05,
      "loss": 0.2698,
      "step": 991,
      "task_loss": 0.2971741855144501
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7313030708561112,
      "compression_loss": 0.0,
      "distillation_loss": 0.10828962922096252,
      "epoch": 0.94,
      "learning_rate": 4.781168668436532e-05,
      "loss": 0.1087,
      "step": 992,
      "task_loss": 0.11210381239652634
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7313307935874732,
      "compression_loss": 0.0,
      "distillation_loss": 0.3468828499317169,
      "epoch": 0.94,
      "learning_rate": 4.780732502660943e-05,
      "loss": 0.3356,
      "step": 993,
      "task_loss": 0.2339598685503006
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7313585088594698,
      "compression_loss": 0.0,
      "distillation_loss": 0.14131498336791992,
      "epoch": 0.94,
      "learning_rate": 4.780295922579312e-05,
      "loss": 0.1407,
      "step": 994,
      "task_loss": 0.1348223239183426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7313862166731049,
      "compression_loss": 0.0,
      "distillation_loss": 0.2651304304599762,
      "epoch": 0.94,
      "learning_rate": 4.779858928270944e-05,
      "loss": 0.253,
      "step": 995,
      "task_loss": 0.1437424123287201
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.731413917029382,
      "compression_loss": 0.0,
      "distillation_loss": 0.11093267053365707,
      "epoch": 0.95,
      "learning_rate": 4.7794215198152216e-05,
      "loss": 0.1031,
      "step": 996,
      "task_loss": 0.03270239382982254
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7314416099293048,
      "compression_loss": 0.0,
      "distillation_loss": 0.2656594514846802,
      "epoch": 0.95,
      "learning_rate": 4.778983697291603e-05,
      "loss": 0.2697,
      "step": 997,
      "task_loss": 0.3064509630203247
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7314692953738771,
      "compression_loss": 0.0,
      "distillation_loss": 0.29231274127960205,
      "epoch": 0.95,
      "learning_rate": 4.7785454607796195e-05,
      "loss": 0.2753,
      "step": 998,
      "task_loss": 0.12249677628278732
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7314969733641024,
      "compression_loss": 0.0,
      "distillation_loss": 0.15542489290237427,
      "epoch": 0.95,
      "learning_rate": 4.77810681035888e-05,
      "loss": 0.154,
      "step": 999,
      "task_loss": 0.14133189618587494
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7315246439009845,
      "compression_loss": 0.0,
      "distillation_loss": 0.27466249465942383,
      "epoch": 0.95,
      "learning_rate": 4.777667746109067e-05,
      "loss": 0.2681,
      "step": 1000,
      "task_loss": 0.20898157358169556
    },
    {
      "epoch": 0.95,
      "eval_accuracy": 0.9094036697247706,
      "eval_loss": 0.3636666536331177,
      "eval_runtime": 17.8028,
      "eval_samples_per_second": 48.981,
      "eval_steps_per_second": 6.123,
      "step": 1000
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7315523069855271,
      "compression_loss": 0.0,
      "distillation_loss": 0.10684286057949066,
      "epoch": 0.95,
      "learning_rate": 4.7772282681099377e-05,
      "loss": 0.1,
      "step": 1001,
      "task_loss": 0.038139708340168
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7315799626187338,
      "compression_loss": 0.0,
      "distillation_loss": 0.33934301137924194,
      "epoch": 0.95,
      "learning_rate": 4.7767883764413266e-05,
      "loss": 0.3332,
      "step": 1002,
      "task_loss": 0.27778148651123047
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7316076108016083,
      "compression_loss": 0.0,
      "distillation_loss": 0.2035553753376007,
      "epoch": 0.95,
      "learning_rate": 4.776348071183142e-05,
      "loss": 0.1945,
      "step": 1003,
      "task_loss": 0.11305206269025803
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7316352515351543,
      "compression_loss": 0.0,
      "distillation_loss": 0.1286626160144806,
      "epoch": 0.95,
      "learning_rate": 4.775907352415367e-05,
      "loss": 0.1426,
      "step": 1004,
      "task_loss": 0.26838886737823486
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7316628848203756,
      "compression_loss": 0.0,
      "distillation_loss": 0.4252236485481262,
      "epoch": 0.95,
      "learning_rate": 4.7754662202180606e-05,
      "loss": 0.4083,
      "step": 1005,
      "task_loss": 0.2558348774909973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7316905106582756,
      "compression_loss": 0.0,
      "distillation_loss": 0.24486252665519714,
      "epoch": 0.96,
      "learning_rate": 4.7750246746713565e-05,
      "loss": 0.2329,
      "step": 1006,
      "task_loss": 0.12567219138145447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7317181290498582,
      "compression_loss": 0.0,
      "distillation_loss": 0.30694228410720825,
      "epoch": 0.96,
      "learning_rate": 4.7745827158554634e-05,
      "loss": 0.2926,
      "step": 1007,
      "task_loss": 0.1638035923242569
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7317457399961271,
      "compression_loss": 0.0,
      "distillation_loss": 0.22996553778648376,
      "epoch": 0.96,
      "learning_rate": 4.774140343850666e-05,
      "loss": 0.227,
      "step": 1008,
      "task_loss": 0.19992922246456146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.731773343498086,
      "compression_loss": 0.0,
      "distillation_loss": 0.17360806465148926,
      "epoch": 0.96,
      "learning_rate": 4.773697558737322e-05,
      "loss": 0.1746,
      "step": 1009,
      "task_loss": 0.1831490695476532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7318009395567384,
      "compression_loss": 0.0,
      "distillation_loss": 0.08699658513069153,
      "epoch": 0.96,
      "learning_rate": 4.773254360595867e-05,
      "loss": 0.082,
      "step": 1010,
      "task_loss": 0.03714621439576149
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7318285281730881,
      "compression_loss": 0.0,
      "distillation_loss": 0.12660613656044006,
      "epoch": 0.96,
      "learning_rate": 4.77281074950681e-05,
      "loss": 0.1187,
      "step": 1011,
      "task_loss": 0.04724828153848648
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7318561093481386,
      "compression_loss": 0.0,
      "distillation_loss": 0.1175093725323677,
      "epoch": 0.96,
      "learning_rate": 4.7723667255507334e-05,
      "loss": 0.1299,
      "step": 1012,
      "task_loss": 0.24159134924411774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.731883683082894,
      "compression_loss": 0.0,
      "distillation_loss": 0.2919718027114868,
      "epoch": 0.96,
      "learning_rate": 4.771922288808297e-05,
      "loss": 0.2807,
      "step": 1013,
      "task_loss": 0.17899443209171295
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7319112493783576,
      "compression_loss": 0.0,
      "distillation_loss": 0.13135388493537903,
      "epoch": 0.96,
      "learning_rate": 4.771477439360235e-05,
      "loss": 0.1246,
      "step": 1014,
      "task_loss": 0.06415353715419769
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7319388082355331,
      "compression_loss": 0.0,
      "distillation_loss": 0.35100582242012024,
      "epoch": 0.96,
      "learning_rate": 4.7710321772873566e-05,
      "loss": 0.348,
      "step": 1015,
      "task_loss": 0.3205385208129883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7319663596554244,
      "compression_loss": 0.0,
      "distillation_loss": 0.2460920214653015,
      "epoch": 0.96,
      "learning_rate": 4.770586502670546e-05,
      "loss": 0.2355,
      "step": 1016,
      "task_loss": 0.13989022374153137
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7319939036390352,
      "compression_loss": 0.0,
      "distillation_loss": 0.23754239082336426,
      "epoch": 0.97,
      "learning_rate": 4.770140415590762e-05,
      "loss": 0.2337,
      "step": 1017,
      "task_loss": 0.19897620379924774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7320214401873689,
      "compression_loss": 0.0,
      "distillation_loss": 0.46301230788230896,
      "epoch": 0.97,
      "learning_rate": 4.769693916129039e-05,
      "loss": 0.4469,
      "step": 1018,
      "task_loss": 0.30206912755966187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7320489693014294,
      "compression_loss": 0.0,
      "distillation_loss": 0.41447192430496216,
      "epoch": 0.97,
      "learning_rate": 4.769247004366485e-05,
      "loss": 0.3952,
      "step": 1019,
      "task_loss": 0.22126147150993347
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7320764909822203,
      "compression_loss": 0.0,
      "distillation_loss": 0.19669479131698608,
      "epoch": 0.97,
      "learning_rate": 4.768799680384283e-05,
      "loss": 0.1961,
      "step": 1020,
      "task_loss": 0.19112606346607208
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7321040052307453,
      "compression_loss": 0.0,
      "distillation_loss": 0.1391059011220932,
      "epoch": 0.97,
      "learning_rate": 4.768351944263693e-05,
      "loss": 0.1306,
      "step": 1021,
      "task_loss": 0.05390199273824692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7321315120480081,
      "compression_loss": 0.0,
      "distillation_loss": 0.1612778604030609,
      "epoch": 0.97,
      "learning_rate": 4.767903796086048e-05,
      "loss": 0.1569,
      "step": 1022,
      "task_loss": 0.11706437915563583
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7321590114350123,
      "compression_loss": 0.0,
      "distillation_loss": 0.13082058727741241,
      "epoch": 0.97,
      "learning_rate": 4.767455235932756e-05,
      "loss": 0.1235,
      "step": 1023,
      "task_loss": 0.05800522491335869
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7321865033927617,
      "compression_loss": 0.0,
      "distillation_loss": 0.07101086527109146,
      "epoch": 0.97,
      "learning_rate": 4.7670062638853e-05,
      "loss": 0.0663,
      "step": 1024,
      "task_loss": 0.023550687357783318
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7322139879222599,
      "compression_loss": 0.0,
      "distillation_loss": 0.7248010635375977,
      "epoch": 0.97,
      "learning_rate": 4.766556880025238e-05,
      "loss": 0.6946,
      "step": 1025,
      "task_loss": 0.4223037362098694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7322414650245106,
      "compression_loss": 0.0,
      "distillation_loss": 0.18530097603797913,
      "epoch": 0.97,
      "learning_rate": 4.7661070844342033e-05,
      "loss": 0.1791,
      "step": 1026,
      "task_loss": 0.12346015125513077
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7322689347005176,
      "compression_loss": 0.0,
      "distillation_loss": 0.436127245426178,
      "epoch": 0.98,
      "learning_rate": 4.7656568771939024e-05,
      "loss": 0.4136,
      "step": 1027,
      "task_loss": 0.21047565340995789
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7322963969512843,
      "compression_loss": 0.0,
      "distillation_loss": 0.33645620942115784,
      "epoch": 0.98,
      "learning_rate": 4.765206258386119e-05,
      "loss": 0.3199,
      "step": 1028,
      "task_loss": 0.17079196870326996
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7323238517778147,
      "compression_loss": 0.0,
      "distillation_loss": 0.2837325930595398,
      "epoch": 0.98,
      "learning_rate": 4.7647552280927086e-05,
      "loss": 0.2713,
      "step": 1029,
      "task_loss": 0.15971429646015167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7323512991811123,
      "compression_loss": 0.0,
      "distillation_loss": 0.2346368283033371,
      "epoch": 0.98,
      "learning_rate": 4.764303786395604e-05,
      "loss": 0.2218,
      "step": 1030,
      "task_loss": 0.10661163181066513
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7323787391621808,
      "compression_loss": 0.0,
      "distillation_loss": 0.26597869396209717,
      "epoch": 0.98,
      "learning_rate": 4.763851933376812e-05,
      "loss": 0.2678,
      "step": 1031,
      "task_loss": 0.284251868724823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7324061717220239,
      "compression_loss": 0.0,
      "distillation_loss": 0.21545451879501343,
      "epoch": 0.98,
      "learning_rate": 4.763399669118414e-05,
      "loss": 0.2012,
      "step": 1032,
      "task_loss": 0.07268868386745453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7324335968616453,
      "compression_loss": 0.0,
      "distillation_loss": 0.3080711364746094,
      "epoch": 0.98,
      "learning_rate": 4.762946993702565e-05,
      "loss": 0.2918,
      "step": 1033,
      "task_loss": 0.14552035927772522
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7324610145820487,
      "compression_loss": 0.0,
      "distillation_loss": 0.2603287100791931,
      "epoch": 0.98,
      "learning_rate": 4.7624939072114954e-05,
      "loss": 0.2473,
      "step": 1034,
      "task_loss": 0.12976713478565216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7324884248842377,
      "compression_loss": 0.0,
      "distillation_loss": 0.28742021322250366,
      "epoch": 0.98,
      "learning_rate": 4.762040409727512e-05,
      "loss": 0.2733,
      "step": 1035,
      "task_loss": 0.14594219624996185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7325158277692161,
      "compression_loss": 0.0,
      "distillation_loss": 0.17579105496406555,
      "epoch": 0.98,
      "learning_rate": 4.761586501332994e-05,
      "loss": 0.1813,
      "step": 1036,
      "task_loss": 0.23123717308044434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7325432232379875,
      "compression_loss": 0.0,
      "distillation_loss": 0.1056813895702362,
      "epoch": 0.98,
      "learning_rate": 4.7611321821103954e-05,
      "loss": 0.1001,
      "step": 1037,
      "task_loss": 0.04944942146539688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7325706112915555,
      "compression_loss": 0.0,
      "distillation_loss": 0.10237106680870056,
      "epoch": 0.99,
      "learning_rate": 4.760677452142247e-05,
      "loss": 0.098,
      "step": 1038,
      "task_loss": 0.05824420601129532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.732597991930924,
      "compression_loss": 0.0,
      "distillation_loss": 0.28087666630744934,
      "epoch": 0.99,
      "learning_rate": 4.760222311511152e-05,
      "loss": 0.273,
      "step": 1039,
      "task_loss": 0.20193740725517273
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7326253651570965,
      "compression_loss": 0.0,
      "distillation_loss": 0.31947529315948486,
      "epoch": 0.99,
      "learning_rate": 4.759766760299788e-05,
      "loss": 0.316,
      "step": 1040,
      "task_loss": 0.2846387028694153
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7326527309710767,
      "compression_loss": 0.0,
      "distillation_loss": 0.08732345700263977,
      "epoch": 0.99,
      "learning_rate": 4.759310798590909e-05,
      "loss": 0.0912,
      "step": 1041,
      "task_loss": 0.12588536739349365
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7326800893738684,
      "compression_loss": 0.0,
      "distillation_loss": 0.31434834003448486,
      "epoch": 0.99,
      "learning_rate": 4.758854426467343e-05,
      "loss": 0.3049,
      "step": 1042,
      "task_loss": 0.22009915113449097
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7327074403664752,
      "compression_loss": 0.0,
      "distillation_loss": 0.3221967816352844,
      "epoch": 0.99,
      "learning_rate": 4.758397644011992e-05,
      "loss": 0.3095,
      "step": 1043,
      "task_loss": 0.19476738572120667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7327347839499008,
      "compression_loss": 0.0,
      "distillation_loss": 0.2775757908821106,
      "epoch": 0.99,
      "learning_rate": 4.757940451307831e-05,
      "loss": 0.2682,
      "step": 1044,
      "task_loss": 0.18372797966003418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7327621201251489,
      "compression_loss": 0.0,
      "distillation_loss": 0.25040361285209656,
      "epoch": 0.99,
      "learning_rate": 4.757482848437914e-05,
      "loss": 0.2413,
      "step": 1045,
      "task_loss": 0.15979930758476257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.732789448893223,
      "compression_loss": 0.0,
      "distillation_loss": 0.16249604523181915,
      "epoch": 0.99,
      "learning_rate": 4.7570248354853644e-05,
      "loss": 0.1538,
      "step": 1046,
      "task_loss": 0.07583259046077728
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.732816770255127,
      "compression_loss": 0.0,
      "distillation_loss": 0.09381724894046783,
      "epoch": 0.99,
      "learning_rate": 4.7565664125333845e-05,
      "loss": 0.0878,
      "step": 1047,
      "task_loss": 0.033302441239356995
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7328440842118646,
      "compression_loss": 0.0,
      "distillation_loss": 0.2704293131828308,
      "epoch": 1.0,
      "learning_rate": 4.7561075796652464e-05,
      "loss": 0.2806,
      "step": 1048,
      "task_loss": 0.3726223111152649
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7328713907644394,
      "compression_loss": 0.0,
      "distillation_loss": 0.2771369516849518,
      "epoch": 1.0,
      "learning_rate": 4.755648336964302e-05,
      "loss": 0.2641,
      "step": 1049,
      "task_loss": 0.14655812084674835
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.732898689913855,
      "compression_loss": 0.0,
      "distillation_loss": 0.11366402357816696,
      "epoch": 1.0,
      "learning_rate": 4.7551886845139743e-05,
      "loss": 0.1051,
      "step": 1050,
      "task_loss": 0.028415286913514137
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7329259816611152,
      "compression_loss": 0.0,
      "distillation_loss": 0.152854323387146,
      "epoch": 1.0,
      "learning_rate": 4.754728622397761e-05,
      "loss": 0.1433,
      "step": 1051,
      "task_loss": 0.05741133540868759
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7329532660072237,
      "compression_loss": 0.0,
      "distillation_loss": 0.18268108367919922,
      "epoch": 1.0,
      "learning_rate": 4.754268150699234e-05,
      "loss": 0.1803,
      "step": 1052,
      "task_loss": 0.15914951264858246
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5468226808906795,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7000000210462276,
      "compression/magnitude_sparsity/target_sparsity_level": 0.732980542953184,
      "compression_loss": 0.0,
      "distillation_loss": 0.04385395348072052,
      "epoch": 1.0,
      "learning_rate": 4.753807269502041e-05,
      "loss": 0.0402,
      "step": 1053,
      "task_loss": 0.007449280470609665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7330078125,
      "compression_loss": 0.0,
      "distillation_loss": 0.29212307929992676,
      "epoch": 1.0,
      "learning_rate": 4.7533459788899026e-05,
      "loss": 0.277,
      "step": 1054,
      "task_loss": 0.1408318132162094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7330350746486752,
      "compression_loss": 0.0,
      "distillation_loss": 0.34965944290161133,
      "epoch": 1.0,
      "learning_rate": 4.752884278946614e-05,
      "loss": 0.3271,
      "step": 1055,
      "task_loss": 0.12405388802289963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7330623294002135,
      "compression_loss": 0.0,
      "distillation_loss": 0.46271616220474243,
      "epoch": 1.0,
      "learning_rate": 4.752422169756048e-05,
      "loss": 0.4489,
      "step": 1056,
      "task_loss": 0.3246670067310333
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7330895767556185,
      "compression_loss": 0.0,
      "distillation_loss": 0.26765865087509155,
      "epoch": 1.0,
      "learning_rate": 4.7519596514021464e-05,
      "loss": 0.2574,
      "step": 1057,
      "task_loss": 0.16510917246341705
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7331168167158937,
      "compression_loss": 0.0,
      "distillation_loss": 0.7084404230117798,
      "epoch": 1.0,
      "learning_rate": 4.751496723968929e-05,
      "loss": 0.6848,
      "step": 1058,
      "task_loss": 0.4719713628292084
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7331440492820429,
      "compression_loss": 0.0,
      "distillation_loss": 0.19782431423664093,
      "epoch": 1.01,
      "learning_rate": 4.751033387540488e-05,
      "loss": 0.1849,
      "step": 1059,
      "task_loss": 0.06854899227619171
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7331712744550699,
      "compression_loss": 0.0,
      "distillation_loss": 0.44147199392318726,
      "epoch": 1.01,
      "learning_rate": 4.7505696422009904e-05,
      "loss": 0.4319,
      "step": 1060,
      "task_loss": 0.34560465812683105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7331984922359782,
      "compression_loss": 0.0,
      "distillation_loss": 0.5233361124992371,
      "epoch": 1.01,
      "learning_rate": 4.750105488034679e-05,
      "loss": 0.4983,
      "step": 1061,
      "task_loss": 0.27257034182548523
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7332257026257717,
      "compression_loss": 0.0,
      "distillation_loss": 0.5197453498840332,
      "epoch": 1.01,
      "learning_rate": 4.749640925125869e-05,
      "loss": 0.5011,
      "step": 1062,
      "task_loss": 0.3330010771751404
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7332529056254539,
      "compression_loss": 0.0,
      "distillation_loss": 0.17582668364048004,
      "epoch": 1.01,
      "learning_rate": 4.749175953558951e-05,
      "loss": 0.1842,
      "step": 1063,
      "task_loss": 0.2600041329860687
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7332801012360284,
      "compression_loss": 0.0,
      "distillation_loss": 0.4421820640563965,
      "epoch": 1.01,
      "learning_rate": 4.748710573418388e-05,
      "loss": 0.4457,
      "step": 1064,
      "task_loss": 0.47691667079925537
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7333072894584991,
      "compression_loss": 0.0,
      "distillation_loss": 0.18625092506408691,
      "epoch": 1.01,
      "learning_rate": 4.7482447847887204e-05,
      "loss": 0.1779,
      "step": 1065,
      "task_loss": 0.1030142530798912
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7333344702938697,
      "compression_loss": 0.0,
      "distillation_loss": 0.19206887483596802,
      "epoch": 1.01,
      "learning_rate": 4.747778587754559e-05,
      "loss": 0.1904,
      "step": 1066,
      "task_loss": 0.1756243109703064
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7333616437431436,
      "compression_loss": 0.0,
      "distillation_loss": 0.28138798475265503,
      "epoch": 1.01,
      "learning_rate": 4.7473119824005926e-05,
      "loss": 0.2826,
      "step": 1067,
      "task_loss": 0.29329320788383484
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7333888098073248,
      "compression_loss": 0.0,
      "distillation_loss": 0.14245599508285522,
      "epoch": 1.01,
      "learning_rate": 4.7468449688115806e-05,
      "loss": 0.1334,
      "step": 1068,
      "task_loss": 0.05212767794728279
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7334159684874167,
      "compression_loss": 0.0,
      "distillation_loss": 0.517392635345459,
      "epoch": 1.02,
      "learning_rate": 4.74637754707236e-05,
      "loss": 0.5164,
      "step": 1069,
      "task_loss": 0.5071411728858948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7334431197844233,
      "compression_loss": 0.0,
      "distillation_loss": 0.3079312741756439,
      "epoch": 1.02,
      "learning_rate": 4.7459097172678386e-05,
      "loss": 0.3028,
      "step": 1070,
      "task_loss": 0.2564443051815033
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.733470263699348,
      "compression_loss": 0.0,
      "distillation_loss": 0.07088293135166168,
      "epoch": 1.02,
      "learning_rate": 4.745441479483001e-05,
      "loss": 0.0657,
      "step": 1071,
      "task_loss": 0.019033435732126236
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7334974002331945,
      "compression_loss": 0.0,
      "distillation_loss": 0.1880199909210205,
      "epoch": 1.02,
      "learning_rate": 4.744972833802904e-05,
      "loss": 0.1756,
      "step": 1072,
      "task_loss": 0.06384404003620148
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7335245293869668,
      "compression_loss": 0.0,
      "distillation_loss": 0.36879584193229675,
      "epoch": 1.02,
      "learning_rate": 4.74450378031268e-05,
      "loss": 0.3632,
      "step": 1073,
      "task_loss": 0.3126252293586731
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7335516511616682,
      "compression_loss": 0.0,
      "distillation_loss": 0.33385515213012695,
      "epoch": 1.02,
      "learning_rate": 4.744034319097535e-05,
      "loss": 0.3191,
      "step": 1074,
      "task_loss": 0.18633373081684113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7335787655583026,
      "compression_loss": 0.0,
      "distillation_loss": 0.3007218539714813,
      "epoch": 1.02,
      "learning_rate": 4.743564450242749e-05,
      "loss": 0.2902,
      "step": 1075,
      "task_loss": 0.19521453976631165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7336058725778736,
      "compression_loss": 0.0,
      "distillation_loss": 0.2639384865760803,
      "epoch": 1.02,
      "learning_rate": 4.7430941738336745e-05,
      "loss": 0.2496,
      "step": 1076,
      "task_loss": 0.120790995657444
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7336329722213848,
      "compression_loss": 0.0,
      "distillation_loss": 0.3540310859680176,
      "epoch": 1.02,
      "learning_rate": 4.742623489955741e-05,
      "loss": 0.3531,
      "step": 1077,
      "task_loss": 0.34429335594177246
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7336600644898401,
      "compression_loss": 0.0,
      "distillation_loss": 0.3709731698036194,
      "epoch": 1.02,
      "learning_rate": 4.74215239869445e-05,
      "loss": 0.3647,
      "step": 1078,
      "task_loss": 0.30800601840019226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.733687149384243,
      "compression_loss": 0.0,
      "distillation_loss": 0.22464478015899658,
      "epoch": 1.02,
      "learning_rate": 4.741680900135377e-05,
      "loss": 0.2118,
      "step": 1079,
      "task_loss": 0.09570850431919098
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7337142269055972,
      "compression_loss": 0.0,
      "distillation_loss": 0.4049076437950134,
      "epoch": 1.03,
      "learning_rate": 4.741208994364173e-05,
      "loss": 0.3865,
      "step": 1080,
      "task_loss": 0.2211461365222931
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7337412970549065,
      "compression_loss": 0.0,
      "distillation_loss": 0.1285635232925415,
      "epoch": 1.03,
      "learning_rate": 4.740736681466561e-05,
      "loss": 0.1211,
      "step": 1081,
      "task_loss": 0.0543033629655838
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7337683598331745,
      "compression_loss": 0.0,
      "distillation_loss": 0.15440712869167328,
      "epoch": 1.03,
      "learning_rate": 4.74026396152834e-05,
      "loss": 0.1622,
      "step": 1082,
      "task_loss": 0.2323606014251709
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7337954152414049,
      "compression_loss": 0.0,
      "distillation_loss": 0.06974273920059204,
      "epoch": 1.03,
      "learning_rate": 4.7397908346353796e-05,
      "loss": 0.0666,
      "step": 1083,
      "task_loss": 0.038624271750450134
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7338224632806014,
      "compression_loss": 0.0,
      "distillation_loss": 0.41569557785987854,
      "epoch": 1.03,
      "learning_rate": 4.739317300873628e-05,
      "loss": 0.3924,
      "step": 1084,
      "task_loss": 0.1831234097480774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7338495039517676,
      "compression_loss": 0.0,
      "distillation_loss": 0.3054138720035553,
      "epoch": 1.03,
      "learning_rate": 4.738843360329104e-05,
      "loss": 0.2951,
      "step": 1085,
      "task_loss": 0.20198288559913635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7338765372559072,
      "compression_loss": 0.0,
      "distillation_loss": 0.26133251190185547,
      "epoch": 1.03,
      "learning_rate": 4.738369013087902e-05,
      "loss": 0.2525,
      "step": 1086,
      "task_loss": 0.17272129654884338
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7339035631940241,
      "compression_loss": 0.0,
      "distillation_loss": 0.20728163421154022,
      "epoch": 1.03,
      "learning_rate": 4.7378942592361876e-05,
      "loss": 0.1958,
      "step": 1087,
      "task_loss": 0.09262159466743469
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7339305817671216,
      "compression_loss": 0.0,
      "distillation_loss": 0.11674150079488754,
      "epoch": 1.03,
      "learning_rate": 4.737419098860204e-05,
      "loss": 0.108,
      "step": 1088,
      "task_loss": 0.029347870498895645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7339575929762037,
      "compression_loss": 0.0,
      "distillation_loss": 0.3030095100402832,
      "epoch": 1.03,
      "learning_rate": 4.7369435320462654e-05,
      "loss": 0.2916,
      "step": 1089,
      "task_loss": 0.1885221302509308
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7339845968222738,
      "compression_loss": 0.0,
      "distillation_loss": 0.05839750915765762,
      "epoch": 1.04,
      "learning_rate": 4.73646755888076e-05,
      "loss": 0.054,
      "step": 1090,
      "task_loss": 0.014137573540210724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7340115933063359,
      "compression_loss": 0.0,
      "distillation_loss": 0.1552124172449112,
      "epoch": 1.04,
      "learning_rate": 4.7359911794501526e-05,
      "loss": 0.1474,
      "step": 1091,
      "task_loss": 0.07671723514795303
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7340385824293935,
      "compression_loss": 0.0,
      "distillation_loss": 0.273028165102005,
      "epoch": 1.04,
      "learning_rate": 4.7355143938409785e-05,
      "loss": 0.2641,
      "step": 1092,
      "task_loss": 0.18373528122901917
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7340655641924503,
      "compression_loss": 0.0,
      "distillation_loss": 0.2551477551460266,
      "epoch": 1.04,
      "learning_rate": 4.735037202139849e-05,
      "loss": 0.2586,
      "step": 1093,
      "task_loss": 0.28983938694000244
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.73409253859651,
      "compression_loss": 0.0,
      "distillation_loss": 0.15983590483665466,
      "epoch": 1.04,
      "learning_rate": 4.734559604433447e-05,
      "loss": 0.1529,
      "step": 1094,
      "task_loss": 0.09036150574684143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7341195056425763,
      "compression_loss": 0.0,
      "distillation_loss": 0.13693635165691376,
      "epoch": 1.04,
      "learning_rate": 4.734081600808531e-05,
      "loss": 0.1312,
      "step": 1095,
      "task_loss": 0.07925444841384888
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7341464653316528,
      "compression_loss": 0.0,
      "distillation_loss": 0.17353419959545135,
      "epoch": 1.04,
      "learning_rate": 4.733603191351933e-05,
      "loss": 0.1626,
      "step": 1096,
      "task_loss": 0.06457867473363876
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7341734176647433,
      "compression_loss": 0.0,
      "distillation_loss": 0.15793973207473755,
      "epoch": 1.04,
      "learning_rate": 4.733124376150558e-05,
      "loss": 0.1539,
      "step": 1097,
      "task_loss": 0.11756959557533264
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7342003626428515,
      "compression_loss": 0.0,
      "distillation_loss": 0.08797385543584824,
      "epoch": 1.04,
      "learning_rate": 4.7326451552913856e-05,
      "loss": 0.0843,
      "step": 1098,
      "task_loss": 0.05155399069190025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7342273002669808,
      "compression_loss": 0.0,
      "distillation_loss": 0.1670214980840683,
      "epoch": 1.04,
      "learning_rate": 4.7321655288614674e-05,
      "loss": 0.1685,
      "step": 1099,
      "task_loss": 0.18175487220287323
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7342542305381352,
      "compression_loss": 0.0,
      "distillation_loss": 0.27932503819465637,
      "epoch": 1.04,
      "learning_rate": 4.7316854969479314e-05,
      "loss": 0.271,
      "step": 1100,
      "task_loss": 0.1957254558801651
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7342811534573183,
      "compression_loss": 0.0,
      "distillation_loss": 0.3590747117996216,
      "epoch": 1.05,
      "learning_rate": 4.7312050596379764e-05,
      "loss": 0.3598,
      "step": 1101,
      "task_loss": 0.36656391620635986
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7343080690255337,
      "compression_loss": 0.0,
      "distillation_loss": 0.17354151606559753,
      "epoch": 1.05,
      "learning_rate": 4.730724217018877e-05,
      "loss": 0.1851,
      "step": 1102,
      "task_loss": 0.2888186275959015
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7343349772437852,
      "compression_loss": 0.0,
      "distillation_loss": 0.17269417643547058,
      "epoch": 1.05,
      "learning_rate": 4.7302429691779806e-05,
      "loss": 0.1714,
      "step": 1103,
      "task_loss": 0.1596398800611496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7343618781130763,
      "compression_loss": 0.0,
      "distillation_loss": 0.23135778307914734,
      "epoch": 1.05,
      "learning_rate": 4.729761316202708e-05,
      "loss": 0.2202,
      "step": 1104,
      "task_loss": 0.11965839564800262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7343887716344109,
      "compression_loss": 0.0,
      "distillation_loss": 0.24613159894943237,
      "epoch": 1.05,
      "learning_rate": 4.729279258180553e-05,
      "loss": 0.2429,
      "step": 1105,
      "task_loss": 0.21396657824516296
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7344156578087925,
      "compression_loss": 0.0,
      "distillation_loss": 0.34451723098754883,
      "epoch": 1.05,
      "learning_rate": 4.7287967951990855e-05,
      "loss": 0.3342,
      "step": 1106,
      "task_loss": 0.2409912496805191
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7344425366372249,
      "compression_loss": 0.0,
      "distillation_loss": 0.15786179900169373,
      "epoch": 1.05,
      "learning_rate": 4.7283139273459445e-05,
      "loss": 0.1565,
      "step": 1107,
      "task_loss": 0.14464277029037476
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7344694081207117,
      "compression_loss": 0.0,
      "distillation_loss": 0.18467701971530914,
      "epoch": 1.05,
      "learning_rate": 4.727830654708848e-05,
      "loss": 0.1777,
      "step": 1108,
      "task_loss": 0.11510906368494034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7344962722602567,
      "compression_loss": 0.0,
      "distillation_loss": 0.22931762039661407,
      "epoch": 1.05,
      "learning_rate": 4.727346977375584e-05,
      "loss": 0.22,
      "step": 1109,
      "task_loss": 0.13630938529968262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7345231290568635,
      "compression_loss": 0.0,
      "distillation_loss": 0.35741716623306274,
      "epoch": 1.05,
      "learning_rate": 4.7268628954340136e-05,
      "loss": 0.3397,
      "step": 1110,
      "task_loss": 0.18016816675662994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7345499785115357,
      "compression_loss": 0.0,
      "distillation_loss": 0.24529217183589935,
      "epoch": 1.06,
      "learning_rate": 4.726378408972074e-05,
      "loss": 0.238,
      "step": 1111,
      "task_loss": 0.1728375256061554
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7345768206252772,
      "compression_loss": 0.0,
      "distillation_loss": 0.29263418912887573,
      "epoch": 1.06,
      "learning_rate": 4.725893518077774e-05,
      "loss": 0.2744,
      "step": 1112,
      "task_loss": 0.11022615432739258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7346036553990914,
      "compression_loss": 0.0,
      "distillation_loss": 0.12653577327728271,
      "epoch": 1.06,
      "learning_rate": 4.725408222839197e-05,
      "loss": 0.1174,
      "step": 1113,
      "task_loss": 0.034942492842674255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7346304828339822,
      "compression_loss": 0.0,
      "distillation_loss": 0.37742096185684204,
      "epoch": 1.06,
      "learning_rate": 4.724922523344498e-05,
      "loss": 0.3659,
      "step": 1114,
      "task_loss": 0.26208043098449707
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7346573029309533,
      "compression_loss": 0.0,
      "distillation_loss": 0.40198618173599243,
      "epoch": 1.06,
      "learning_rate": 4.724436419681907e-05,
      "loss": 0.3978,
      "step": 1115,
      "task_loss": 0.3604857623577118
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7346841156910082,
      "compression_loss": 0.0,
      "distillation_loss": 0.24844889342784882,
      "epoch": 1.06,
      "learning_rate": 4.723949911939728e-05,
      "loss": 0.2425,
      "step": 1116,
      "task_loss": 0.1891540288925171
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7347109211151508,
      "compression_loss": 0.0,
      "distillation_loss": 0.5047147274017334,
      "epoch": 1.06,
      "learning_rate": 4.723463000206337e-05,
      "loss": 0.4934,
      "step": 1117,
      "task_loss": 0.39201080799102783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7347377192043846,
      "compression_loss": 0.0,
      "distillation_loss": 0.1238306388258934,
      "epoch": 1.06,
      "learning_rate": 4.722975684570183e-05,
      "loss": 0.1146,
      "step": 1118,
      "task_loss": 0.03139756619930267
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7347645099597133,
      "compression_loss": 0.0,
      "distillation_loss": 0.23237791657447815,
      "epoch": 1.06,
      "learning_rate": 4.7224879651197905e-05,
      "loss": 0.2398,
      "step": 1119,
      "task_loss": 0.30690667033195496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7347912933821407,
      "compression_loss": 0.0,
      "distillation_loss": 0.20782580971717834,
      "epoch": 1.06,
      "learning_rate": 4.721999841943755e-05,
      "loss": 0.2096,
      "step": 1120,
      "task_loss": 0.22543630003929138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7348180694726704,
      "compression_loss": 0.0,
      "distillation_loss": 0.18192127346992493,
      "epoch": 1.06,
      "learning_rate": 4.721511315130747e-05,
      "loss": 0.1793,
      "step": 1121,
      "task_loss": 0.15596555173397064
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7348448382323061,
      "compression_loss": 0.0,
      "distillation_loss": 0.14186906814575195,
      "epoch": 1.07,
      "learning_rate": 4.7210223847695104e-05,
      "loss": 0.1417,
      "step": 1122,
      "task_loss": 0.14012937247753143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7348715996620514,
      "compression_loss": 0.0,
      "distillation_loss": 0.33551573753356934,
      "epoch": 1.07,
      "learning_rate": 4.72053305094886e-05,
      "loss": 0.3328,
      "step": 1123,
      "task_loss": 0.3087965250015259
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.73489835376291,
      "compression_loss": 0.0,
      "distillation_loss": 0.16018086671829224,
      "epoch": 1.07,
      "learning_rate": 4.720043313757687e-05,
      "loss": 0.1579,
      "step": 1124,
      "task_loss": 0.13755735754966736
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7349251005358858,
      "compression_loss": 0.0,
      "distillation_loss": 0.0789952278137207,
      "epoch": 1.07,
      "learning_rate": 4.719553173284955e-05,
      "loss": 0.0733,
      "step": 1125,
      "task_loss": 0.021938461810350418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7349518399819822,
      "compression_loss": 0.0,
      "distillation_loss": 0.1255275011062622,
      "epoch": 1.07,
      "learning_rate": 4.719062629619699e-05,
      "loss": 0.1193,
      "step": 1126,
      "task_loss": 0.06374667584896088
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7349785721022031,
      "compression_loss": 0.0,
      "distillation_loss": 0.14572109282016754,
      "epoch": 1.07,
      "learning_rate": 4.71857168285103e-05,
      "loss": 0.1458,
      "step": 1127,
      "task_loss": 0.14671580493450165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.735005296897552,
      "compression_loss": 0.0,
      "distillation_loss": 0.18587951362133026,
      "epoch": 1.07,
      "learning_rate": 4.718080333068129e-05,
      "loss": 0.186,
      "step": 1128,
      "task_loss": 0.18705090880393982
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7350320143690328,
      "compression_loss": 0.0,
      "distillation_loss": 0.31489160656929016,
      "epoch": 1.07,
      "learning_rate": 4.717588580360253e-05,
      "loss": 0.2983,
      "step": 1129,
      "task_loss": 0.14935556054115295
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7350587245176489,
      "compression_loss": 0.0,
      "distillation_loss": 0.2369147539138794,
      "epoch": 1.07,
      "learning_rate": 4.717096424816731e-05,
      "loss": 0.2477,
      "step": 1130,
      "task_loss": 0.3448143005371094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7350854273444042,
      "compression_loss": 0.0,
      "distillation_loss": 0.24669213593006134,
      "epoch": 1.07,
      "learning_rate": 4.716603866526967e-05,
      "loss": 0.2424,
      "step": 1131,
      "task_loss": 0.20330274105072021
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7351121228503023,
      "compression_loss": 0.0,
      "distillation_loss": 0.29487940669059753,
      "epoch": 1.08,
      "learning_rate": 4.7161109055804356e-05,
      "loss": 0.286,
      "step": 1132,
      "task_loss": 0.20643703639507294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7351388110363469,
      "compression_loss": 0.0,
      "distillation_loss": 0.35977235436439514,
      "epoch": 1.08,
      "learning_rate": 4.7156175420666844e-05,
      "loss": 0.3461,
      "step": 1133,
      "task_loss": 0.22308805584907532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7351654919035416,
      "compression_loss": 0.0,
      "distillation_loss": 0.21201397478580475,
      "epoch": 1.08,
      "learning_rate": 4.715123776075336e-05,
      "loss": 0.2005,
      "step": 1134,
      "task_loss": 0.09711904078722
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7351921654528903,
      "compression_loss": 0.0,
      "distillation_loss": 0.16644607484340668,
      "epoch": 1.08,
      "learning_rate": 4.714629607696086e-05,
      "loss": 0.1694,
      "step": 1135,
      "task_loss": 0.19630329310894012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7352188316853965,
      "compression_loss": 0.0,
      "distillation_loss": 0.06290404498577118,
      "epoch": 1.08,
      "learning_rate": 4.714135037018702e-05,
      "loss": 0.0586,
      "step": 1136,
      "task_loss": 0.01941380277276039
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7352454906020639,
      "compression_loss": 0.0,
      "distillation_loss": 0.2712949216365814,
      "epoch": 1.08,
      "learning_rate": 4.713640064133025e-05,
      "loss": 0.2624,
      "step": 1137,
      "task_loss": 0.18186871707439423
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7352721422038961,
      "compression_loss": 0.0,
      "distillation_loss": 0.11187370866537094,
      "epoch": 1.08,
      "learning_rate": 4.7131446891289694e-05,
      "loss": 0.1134,
      "step": 1138,
      "task_loss": 0.12678144872188568
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7352987864918971,
      "compression_loss": 0.0,
      "distillation_loss": 0.1483362764120102,
      "epoch": 1.08,
      "learning_rate": 4.712648912096522e-05,
      "loss": 0.1437,
      "step": 1139,
      "task_loss": 0.10216772556304932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7353254234670703,
      "compression_loss": 0.0,
      "distillation_loss": 0.204280823469162,
      "epoch": 1.08,
      "learning_rate": 4.712152733125744e-05,
      "loss": 0.2035,
      "step": 1140,
      "task_loss": 0.196111261844635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7353520531304194,
      "compression_loss": 0.0,
      "distillation_loss": 0.18241867423057556,
      "epoch": 1.08,
      "learning_rate": 4.711656152306768e-05,
      "loss": 0.1709,
      "step": 1141,
      "task_loss": 0.06716363877058029
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7353786754829481,
      "compression_loss": 0.0,
      "distillation_loss": 0.18135663866996765,
      "epoch": 1.08,
      "learning_rate": 4.711159169729801e-05,
      "loss": 0.1737,
      "step": 1142,
      "task_loss": 0.10434912145137787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7354052905256602,
      "compression_loss": 0.0,
      "distillation_loss": 0.16455084085464478,
      "epoch": 1.09,
      "learning_rate": 4.710661785485121e-05,
      "loss": 0.1577,
      "step": 1143,
      "task_loss": 0.09606322646141052
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7354318982595592,
      "compression_loss": 0.0,
      "distillation_loss": 0.18896543979644775,
      "epoch": 1.09,
      "learning_rate": 4.710163999663081e-05,
      "loss": 0.1757,
      "step": 1144,
      "task_loss": 0.05640107020735741
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7354584986856489,
      "compression_loss": 0.0,
      "distillation_loss": 0.21004149317741394,
      "epoch": 1.09,
      "learning_rate": 4.709665812354107e-05,
      "loss": 0.2006,
      "step": 1145,
      "task_loss": 0.11531868577003479
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7354850918049332,
      "compression_loss": 0.0,
      "distillation_loss": 0.06717594712972641,
      "epoch": 1.09,
      "learning_rate": 4.709167223648695e-05,
      "loss": 0.0618,
      "step": 1146,
      "task_loss": 0.013451127335429192
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7355116776184153,
      "compression_loss": 0.0,
      "distillation_loss": 0.06622839719057083,
      "epoch": 1.09,
      "learning_rate": 4.7086682336374187e-05,
      "loss": 0.0771,
      "step": 1147,
      "task_loss": 0.17544050514698029
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7355382561270991,
      "compression_loss": 0.0,
      "distillation_loss": 0.42941948771476746,
      "epoch": 1.09,
      "learning_rate": 4.70816884241092e-05,
      "loss": 0.4142,
      "step": 1148,
      "task_loss": 0.2768740653991699
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7355648273319885,
      "compression_loss": 0.0,
      "distillation_loss": 0.3746296763420105,
      "epoch": 1.09,
      "learning_rate": 4.7076690500599164e-05,
      "loss": 0.3572,
      "step": 1149,
      "task_loss": 0.20025447010993958
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7355913912340868,
      "compression_loss": 0.0,
      "distillation_loss": 0.1592552214860916,
      "epoch": 1.09,
      "learning_rate": 4.707168856675198e-05,
      "loss": 0.1792,
      "step": 1150,
      "task_loss": 0.3586212992668152
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.735617947834398,
      "compression_loss": 0.0,
      "distillation_loss": 0.16237607598304749,
      "epoch": 1.09,
      "learning_rate": 4.7066682623476265e-05,
      "loss": 0.1536,
      "step": 1151,
      "task_loss": 0.07469609379768372
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7356444971339255,
      "compression_loss": 0.0,
      "distillation_loss": 0.18786782026290894,
      "epoch": 1.09,
      "learning_rate": 4.706167267168138e-05,
      "loss": 0.1743,
      "step": 1152,
      "task_loss": 0.05179518088698387
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7356710391336734,
      "compression_loss": 0.0,
      "distillation_loss": 0.41019895672798157,
      "epoch": 1.09,
      "learning_rate": 4.70566587122774e-05,
      "loss": 0.4003,
      "step": 1153,
      "task_loss": 0.311697781085968
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.735697573834645,
      "compression_loss": 0.0,
      "distillation_loss": 0.27642178535461426,
      "epoch": 1.1,
      "learning_rate": 4.7051640746175147e-05,
      "loss": 0.2717,
      "step": 1154,
      "task_loss": 0.22955429553985596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.735724101237844,
      "compression_loss": 0.0,
      "distillation_loss": 0.41415461897850037,
      "epoch": 1.1,
      "learning_rate": 4.7046618774286146e-05,
      "loss": 0.3986,
      "step": 1155,
      "task_loss": 0.2587291896343231
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7357506213442743,
      "compression_loss": 0.0,
      "distillation_loss": 0.10836180299520493,
      "epoch": 1.1,
      "learning_rate": 4.7041592797522664e-05,
      "loss": 0.1089,
      "step": 1156,
      "task_loss": 0.11396181583404541
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7357771341549394,
      "compression_loss": 0.0,
      "distillation_loss": 0.3199521601200104,
      "epoch": 1.1,
      "learning_rate": 4.7036562816797705e-05,
      "loss": 0.3084,
      "step": 1157,
      "task_loss": 0.2041257917881012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7358036396708431,
      "compression_loss": 0.0,
      "distillation_loss": 0.26940402388572693,
      "epoch": 1.1,
      "learning_rate": 4.7031528833024976e-05,
      "loss": 0.2594,
      "step": 1158,
      "task_loss": 0.16965427994728088
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7358301378929889,
      "compression_loss": 0.0,
      "distillation_loss": 0.15760007500648499,
      "epoch": 1.1,
      "learning_rate": 4.702649084711892e-05,
      "loss": 0.167,
      "step": 1159,
      "task_loss": 0.2519673705101013
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7358566288223808,
      "compression_loss": 0.0,
      "distillation_loss": 0.44500085711479187,
      "epoch": 1.1,
      "learning_rate": 4.7021448859994735e-05,
      "loss": 0.4191,
      "step": 1160,
      "task_loss": 0.1861766129732132
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7358831124600221,
      "compression_loss": 0.0,
      "distillation_loss": 0.10013365000486374,
      "epoch": 1.1,
      "learning_rate": 4.70164028725683e-05,
      "loss": 0.093,
      "step": 1161,
      "task_loss": 0.029256567358970642
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7359095888069168,
      "compression_loss": 0.0,
      "distillation_loss": 0.30927249789237976,
      "epoch": 1.1,
      "learning_rate": 4.7011352885756255e-05,
      "loss": 0.2942,
      "step": 1162,
      "task_loss": 0.15870612859725952
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7359360578640685,
      "compression_loss": 0.0,
      "distillation_loss": 0.13697347044944763,
      "epoch": 1.1,
      "learning_rate": 4.7006298900475954e-05,
      "loss": 0.1302,
      "step": 1163,
      "task_loss": 0.06875795125961304
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7359625196324807,
      "compression_loss": 0.0,
      "distillation_loss": 0.37151408195495605,
      "epoch": 1.11,
      "learning_rate": 4.7001240917645465e-05,
      "loss": 0.3618,
      "step": 1164,
      "task_loss": 0.27404022216796875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7359889741131573,
      "compression_loss": 0.0,
      "distillation_loss": 0.08429434895515442,
      "epoch": 1.11,
      "learning_rate": 4.699617893818361e-05,
      "loss": 0.079,
      "step": 1165,
      "task_loss": 0.031680069863796234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.736015421307102,
      "compression_loss": 0.0,
      "distillation_loss": 0.20043613016605377,
      "epoch": 1.11,
      "learning_rate": 4.699111296300992e-05,
      "loss": 0.1903,
      "step": 1166,
      "task_loss": 0.09880296885967255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7360418612153182,
      "compression_loss": 0.0,
      "distillation_loss": 0.3577537536621094,
      "epoch": 1.11,
      "learning_rate": 4.6986042993044645e-05,
      "loss": 0.344,
      "step": 1167,
      "task_loss": 0.220191091299057
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7360682938388099,
      "compression_loss": 0.0,
      "distillation_loss": 0.3422430455684662,
      "epoch": 1.11,
      "learning_rate": 4.698096902920877e-05,
      "loss": 0.3224,
      "step": 1168,
      "task_loss": 0.1436607986688614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7360947191785806,
      "compression_loss": 0.0,
      "distillation_loss": 0.20955124497413635,
      "epoch": 1.11,
      "learning_rate": 4.6975891072424015e-05,
      "loss": 0.2027,
      "step": 1169,
      "task_loss": 0.14070504903793335
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7361211372356341,
      "compression_loss": 0.0,
      "distillation_loss": 0.16709572076797485,
      "epoch": 1.11,
      "learning_rate": 4.697080912361281e-05,
      "loss": 0.1657,
      "step": 1170,
      "task_loss": 0.15344390273094177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7361475480109739,
      "compression_loss": 0.0,
      "distillation_loss": 0.1173921525478363,
      "epoch": 1.11,
      "learning_rate": 4.696572318369831e-05,
      "loss": 0.1092,
      "step": 1171,
      "task_loss": 0.035038650035858154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7361739515056039,
      "compression_loss": 0.0,
      "distillation_loss": 0.09274931252002716,
      "epoch": 1.11,
      "learning_rate": 4.696063325360441e-05,
      "loss": 0.0872,
      "step": 1172,
      "task_loss": 0.03688472509384155
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7362003477205277,
      "compression_loss": 0.0,
      "distillation_loss": 0.19270846247673035,
      "epoch": 1.11,
      "learning_rate": 4.6955539334255716e-05,
      "loss": 0.1832,
      "step": 1173,
      "task_loss": 0.09803463518619537
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7362267366567488,
      "compression_loss": 0.0,
      "distillation_loss": 0.07251520454883575,
      "epoch": 1.11,
      "learning_rate": 4.6950441426577565e-05,
      "loss": 0.0787,
      "step": 1174,
      "task_loss": 0.13475212454795837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7362531183152712,
      "compression_loss": 0.0,
      "distillation_loss": 0.3601876199245453,
      "epoch": 1.12,
      "learning_rate": 4.694533953149601e-05,
      "loss": 0.3481,
      "step": 1175,
      "task_loss": 0.23886962234973907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7362794926970984,
      "compression_loss": 0.0,
      "distillation_loss": 0.24627402424812317,
      "epoch": 1.12,
      "learning_rate": 4.694023364993784e-05,
      "loss": 0.2416,
      "step": 1176,
      "task_loss": 0.19923092424869537
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.736305859803234,
      "compression_loss": 0.0,
      "distillation_loss": 0.1757299304008484,
      "epoch": 1.12,
      "learning_rate": 4.693512378283056e-05,
      "loss": 0.1673,
      "step": 1177,
      "task_loss": 0.09126180410385132
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7363322196346819,
      "compression_loss": 0.0,
      "distillation_loss": 0.127120703458786,
      "epoch": 1.12,
      "learning_rate": 4.693000993110241e-05,
      "loss": 0.1282,
      "step": 1178,
      "task_loss": 0.13783586025238037
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7363585721924456,
      "compression_loss": 0.0,
      "distillation_loss": 0.02453036420047283,
      "epoch": 1.12,
      "learning_rate": 4.692489209568234e-05,
      "loss": 0.0399,
      "step": 1179,
      "task_loss": 0.17848730087280273
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7363849174775289,
      "compression_loss": 0.0,
      "distillation_loss": 0.21834155917167664,
      "epoch": 1.12,
      "learning_rate": 4.691977027750002e-05,
      "loss": 0.2075,
      "step": 1180,
      "task_loss": 0.11009831726551056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7364112554909354,
      "compression_loss": 0.0,
      "distillation_loss": 0.33225932717323303,
      "epoch": 1.12,
      "learning_rate": 4.691464447748587e-05,
      "loss": 0.3258,
      "step": 1181,
      "task_loss": 0.2678440511226654
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7364375862336688,
      "compression_loss": 0.0,
      "distillation_loss": 0.08998244255781174,
      "epoch": 1.12,
      "learning_rate": 4.690951469657101e-05,
      "loss": 0.1002,
      "step": 1182,
      "task_loss": 0.1922500729560852
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7364639097067329,
      "compression_loss": 0.0,
      "distillation_loss": 0.07570269703865051,
      "epoch": 1.12,
      "learning_rate": 4.690438093568728e-05,
      "loss": 0.071,
      "step": 1183,
      "task_loss": 0.028192538768053055
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7364902259111312,
      "compression_loss": 0.0,
      "distillation_loss": 0.1313639134168625,
      "epoch": 1.12,
      "learning_rate": 4.689924319576727e-05,
      "loss": 0.139,
      "step": 1184,
      "task_loss": 0.20730452239513397
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7365165348478675,
      "compression_loss": 0.0,
      "distillation_loss": 0.2448119819164276,
      "epoch": 1.13,
      "learning_rate": 4.689410147774426e-05,
      "loss": 0.2476,
      "step": 1185,
      "task_loss": 0.2722412347793579
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7365428365179455,
      "compression_loss": 0.0,
      "distillation_loss": 0.35139238834381104,
      "epoch": 1.13,
      "learning_rate": 4.6888955782552274e-05,
      "loss": 0.3343,
      "step": 1186,
      "task_loss": 0.18060128390789032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7365691309223688,
      "compression_loss": 0.0,
      "distillation_loss": 0.09512563794851303,
      "epoch": 1.13,
      "learning_rate": 4.688380611112605e-05,
      "loss": 0.0879,
      "step": 1187,
      "task_loss": 0.022444602102041245
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.736595418062141,
      "compression_loss": 0.0,
      "distillation_loss": 0.23818418383598328,
      "epoch": 1.13,
      "learning_rate": 4.687865246440106e-05,
      "loss": 0.2262,
      "step": 1188,
      "task_loss": 0.11839476227760315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7366216979382659,
      "compression_loss": 0.0,
      "distillation_loss": 0.23918560147285461,
      "epoch": 1.13,
      "learning_rate": 4.687349484331347e-05,
      "loss": 0.2253,
      "step": 1189,
      "task_loss": 0.10081011056900024
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7366479705517474,
      "compression_loss": 0.0,
      "distillation_loss": 0.0738721638917923,
      "epoch": 1.13,
      "learning_rate": 4.6868333248800204e-05,
      "loss": 0.0835,
      "step": 1190,
      "task_loss": 0.1698300540447235
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7366742359035887,
      "compression_loss": 0.0,
      "distillation_loss": 0.32272762060165405,
      "epoch": 1.13,
      "learning_rate": 4.686316768179889e-05,
      "loss": 0.3285,
      "step": 1191,
      "task_loss": 0.3799636662006378
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7367004939947939,
      "compression_loss": 0.0,
      "distillation_loss": 0.2319181114435196,
      "epoch": 1.13,
      "learning_rate": 4.685799814324786e-05,
      "loss": 0.2353,
      "step": 1192,
      "task_loss": 0.26583027839660645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7367267448263665,
      "compression_loss": 0.0,
      "distillation_loss": 0.17839057743549347,
      "epoch": 1.13,
      "learning_rate": 4.685282463408621e-05,
      "loss": 0.1696,
      "step": 1193,
      "task_loss": 0.09003418684005737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7367529883993101,
      "compression_loss": 0.0,
      "distillation_loss": 0.05834461748600006,
      "epoch": 1.13,
      "learning_rate": 4.6847647155253716e-05,
      "loss": 0.054,
      "step": 1194,
      "task_loss": 0.014876075088977814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7367792247146285,
      "compression_loss": 0.0,
      "distillation_loss": 0.09124276041984558,
      "epoch": 1.13,
      "learning_rate": 4.684246570769089e-05,
      "loss": 0.0846,
      "step": 1195,
      "task_loss": 0.02449551224708557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7368054537733254,
      "compression_loss": 0.0,
      "distillation_loss": 0.21645143628120422,
      "epoch": 1.14,
      "learning_rate": 4.683728029233898e-05,
      "loss": 0.204,
      "step": 1196,
      "task_loss": 0.09219798445701599
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7368316755764045,
      "compression_loss": 0.0,
      "distillation_loss": 0.062870554625988,
      "epoch": 1.14,
      "learning_rate": 4.683209091013994e-05,
      "loss": 0.0685,
      "step": 1197,
      "task_loss": 0.11939730495214462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7368578901248695,
      "compression_loss": 0.0,
      "distillation_loss": 0.04577638953924179,
      "epoch": 1.14,
      "learning_rate": 4.682689756203643e-05,
      "loss": 0.0508,
      "step": 1198,
      "task_loss": 0.0960756242275238
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7368840974197238,
      "compression_loss": 0.0,
      "distillation_loss": 0.08127100765705109,
      "epoch": 1.14,
      "learning_rate": 4.682170024897187e-05,
      "loss": 0.0913,
      "step": 1199,
      "task_loss": 0.1819191575050354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7369102974619715,
      "compression_loss": 0.0,
      "distillation_loss": 0.06385953724384308,
      "epoch": 1.14,
      "learning_rate": 4.681649897189036e-05,
      "loss": 0.0621,
      "step": 1200,
      "task_loss": 0.04620221257209778
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7369364902526159,
      "compression_loss": 0.0,
      "distillation_loss": 0.3319193720817566,
      "epoch": 1.14,
      "learning_rate": 4.681129373173674e-05,
      "loss": 0.3153,
      "step": 1201,
      "task_loss": 0.16523145139217377
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.736962675792661,
      "compression_loss": 0.0,
      "distillation_loss": 0.11838692426681519,
      "epoch": 1.14,
      "learning_rate": 4.6806084529456574e-05,
      "loss": 0.1378,
      "step": 1202,
      "task_loss": 0.3123231530189514
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7369888540831102,
      "compression_loss": 0.0,
      "distillation_loss": 0.27343183755874634,
      "epoch": 1.14,
      "learning_rate": 4.6800871365996135e-05,
      "loss": 0.2706,
      "step": 1203,
      "task_loss": 0.244797021150589
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7370150251249674,
      "compression_loss": 0.0,
      "distillation_loss": 0.37955009937286377,
      "epoch": 1.14,
      "learning_rate": 4.679565424230241e-05,
      "loss": 0.3633,
      "step": 1204,
      "task_loss": 0.2169659435749054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7370411889192363,
      "compression_loss": 0.0,
      "distillation_loss": 0.11844378709793091,
      "epoch": 1.14,
      "learning_rate": 4.679043315932313e-05,
      "loss": 0.1225,
      "step": 1205,
      "task_loss": 0.15944825112819672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7370673454669204,
      "compression_loss": 0.0,
      "distillation_loss": 0.20653195679187775,
      "epoch": 1.15,
      "learning_rate": 4.6785208118006715e-05,
      "loss": 0.1989,
      "step": 1206,
      "task_loss": 0.1306045949459076
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7370934947690234,
      "compression_loss": 0.0,
      "distillation_loss": 0.16965758800506592,
      "epoch": 1.15,
      "learning_rate": 4.677997911930234e-05,
      "loss": 0.1598,
      "step": 1207,
      "task_loss": 0.07105232030153275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7371196368265492,
      "compression_loss": 0.0,
      "distillation_loss": 0.3983684182167053,
      "epoch": 1.15,
      "learning_rate": 4.6774746164159854e-05,
      "loss": 0.3717,
      "step": 1208,
      "task_loss": 0.13173796236515045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7371457716405012,
      "compression_loss": 0.0,
      "distillation_loss": 0.12735429406166077,
      "epoch": 1.15,
      "learning_rate": 4.676950925352986e-05,
      "loss": 0.1323,
      "step": 1209,
      "task_loss": 0.17691615223884583
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7371718992118833,
      "compression_loss": 0.0,
      "distillation_loss": 0.05615377053618431,
      "epoch": 1.15,
      "learning_rate": 4.676426838836367e-05,
      "loss": 0.054,
      "step": 1210,
      "task_loss": 0.03485414758324623
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7371980195416992,
      "compression_loss": 0.0,
      "distillation_loss": 0.10495641827583313,
      "epoch": 1.15,
      "learning_rate": 4.675902356961331e-05,
      "loss": 0.1051,
      "step": 1211,
      "task_loss": 0.1060037836432457
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7372241326309523,
      "compression_loss": 0.0,
      "distillation_loss": 0.04488696902990341,
      "epoch": 1.15,
      "learning_rate": 4.675377479823153e-05,
      "loss": 0.0409,
      "step": 1212,
      "task_loss": 0.00522448867559433
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7372502384806465,
      "compression_loss": 0.0,
      "distillation_loss": 0.15110138058662415,
      "epoch": 1.15,
      "learning_rate": 4.6748522075171784e-05,
      "loss": 0.1397,
      "step": 1213,
      "task_loss": 0.03717401623725891
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7372763370917854,
      "compression_loss": 0.0,
      "distillation_loss": 0.2912394404411316,
      "epoch": 1.15,
      "learning_rate": 4.674326540138826e-05,
      "loss": 0.2817,
      "step": 1214,
      "task_loss": 0.19608113169670105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7373024284653729,
      "compression_loss": 0.0,
      "distillation_loss": 0.21200403571128845,
      "epoch": 1.15,
      "learning_rate": 4.673800477783587e-05,
      "loss": 0.2112,
      "step": 1215,
      "task_loss": 0.20386581122875214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7373285126024124,
      "compression_loss": 0.0,
      "distillation_loss": 0.29864805936813354,
      "epoch": 1.15,
      "learning_rate": 4.6732740205470206e-05,
      "loss": 0.2818,
      "step": 1216,
      "task_loss": 0.13044750690460205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7373545895039076,
      "compression_loss": 0.0,
      "distillation_loss": 0.12389582395553589,
      "epoch": 1.16,
      "learning_rate": 4.672747168524762e-05,
      "loss": 0.1253,
      "step": 1217,
      "task_loss": 0.13781008124351501
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7373806591708625,
      "compression_loss": 0.0,
      "distillation_loss": 0.04061355069279671,
      "epoch": 1.16,
      "learning_rate": 4.672219921812517e-05,
      "loss": 0.0385,
      "step": 1218,
      "task_loss": 0.019226521253585815
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7374067216042803,
      "compression_loss": 0.0,
      "distillation_loss": 0.11448079347610474,
      "epoch": 1.16,
      "learning_rate": 4.671692280506061e-05,
      "loss": 0.128,
      "step": 1219,
      "task_loss": 0.2497006058692932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7374327768051652,
      "compression_loss": 0.0,
      "distillation_loss": 0.09137353301048279,
      "epoch": 1.16,
      "learning_rate": 4.671164244701243e-05,
      "loss": 0.0936,
      "step": 1220,
      "task_loss": 0.11410032212734222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7374588247745204,
      "compression_loss": 0.0,
      "distillation_loss": 0.11541637778282166,
      "epoch": 1.16,
      "learning_rate": 4.670635814493984e-05,
      "loss": 0.1065,
      "step": 1221,
      "task_loss": 0.025996150448918343
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7374848655133499,
      "compression_loss": 0.0,
      "distillation_loss": 0.0969296246767044,
      "epoch": 1.16,
      "learning_rate": 4.6701069899802755e-05,
      "loss": 0.097,
      "step": 1222,
      "task_loss": 0.09771576523780823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7375108990226572,
      "compression_loss": 0.0,
      "distillation_loss": 0.2986716032028198,
      "epoch": 1.16,
      "learning_rate": 4.669577771256181e-05,
      "loss": 0.2872,
      "step": 1223,
      "task_loss": 0.18374918401241302
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7375369253034462,
      "compression_loss": 0.0,
      "distillation_loss": 0.06657759100198746,
      "epoch": 1.16,
      "learning_rate": 4.6690481584178354e-05,
      "loss": 0.0615,
      "step": 1224,
      "task_loss": 0.015912499278783798
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7375629443567204,
      "compression_loss": 0.0,
      "distillation_loss": 0.20490974187850952,
      "epoch": 1.16,
      "learning_rate": 4.6685181515614454e-05,
      "loss": 0.2002,
      "step": 1225,
      "task_loss": 0.15764397382736206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7375889561834835,
      "compression_loss": 0.0,
      "distillation_loss": 0.12317511439323425,
      "epoch": 1.16,
      "learning_rate": 4.6679877507832895e-05,
      "loss": 0.1218,
      "step": 1226,
      "task_loss": 0.10897918045520782
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7376149607847392,
      "compression_loss": 0.0,
      "distillation_loss": 0.0427418127655983,
      "epoch": 1.17,
      "learning_rate": 4.6674569561797174e-05,
      "loss": 0.0424,
      "step": 1227,
      "task_loss": 0.0389644056558609
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7376409581614912,
      "compression_loss": 0.0,
      "distillation_loss": 0.14768566191196442,
      "epoch": 1.17,
      "learning_rate": 4.666925767847151e-05,
      "loss": 0.1529,
      "step": 1228,
      "task_loss": 0.20031431317329407
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7376669483147432,
      "compression_loss": 0.0,
      "distillation_loss": 0.09107698500156403,
      "epoch": 1.17,
      "learning_rate": 4.6663941858820825e-05,
      "loss": 0.0864,
      "step": 1229,
      "task_loss": 0.04394202679395676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7376929312454988,
      "compression_loss": 0.0,
      "distillation_loss": 0.26698851585388184,
      "epoch": 1.17,
      "learning_rate": 4.665862210381077e-05,
      "loss": 0.2655,
      "step": 1230,
      "task_loss": 0.2520076036453247
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7377189069547618,
      "compression_loss": 0.0,
      "distillation_loss": 0.12991863489151,
      "epoch": 1.17,
      "learning_rate": 4.66532984144077e-05,
      "loss": 0.1312,
      "step": 1231,
      "task_loss": 0.14297693967819214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7377448754435358,
      "compression_loss": 0.0,
      "distillation_loss": 0.22150184214115143,
      "epoch": 1.17,
      "learning_rate": 4.6647970791578685e-05,
      "loss": 0.223,
      "step": 1232,
      "task_loss": 0.23616032302379608
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7377708367128246,
      "compression_loss": 0.0,
      "distillation_loss": 0.32014334201812744,
      "epoch": 1.17,
      "learning_rate": 4.664263923629153e-05,
      "loss": 0.3188,
      "step": 1233,
      "task_loss": 0.30650225281715393
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7377967907636317,
      "compression_loss": 0.0,
      "distillation_loss": 0.485608845949173,
      "epoch": 1.17,
      "learning_rate": 4.663730374951472e-05,
      "loss": 0.4772,
      "step": 1234,
      "task_loss": 0.4013437032699585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7378227375969609,
      "compression_loss": 0.0,
      "distillation_loss": 0.06522876024246216,
      "epoch": 1.17,
      "learning_rate": 4.663196433221747e-05,
      "loss": 0.0715,
      "step": 1235,
      "task_loss": 0.12773452699184418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7378486772138159,
      "compression_loss": 0.0,
      "distillation_loss": 0.2558850646018982,
      "epoch": 1.17,
      "learning_rate": 4.6626620985369724e-05,
      "loss": 0.2443,
      "step": 1236,
      "task_loss": 0.13986220955848694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7378746096152002,
      "compression_loss": 0.0,
      "distillation_loss": 0.07191958278417587,
      "epoch": 1.17,
      "learning_rate": 4.662127370994212e-05,
      "loss": 0.076,
      "step": 1237,
      "task_loss": 0.11259147524833679
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7379005348021177,
      "compression_loss": 0.0,
      "distillation_loss": 0.15925487875938416,
      "epoch": 1.18,
      "learning_rate": 4.6615922506906016e-05,
      "loss": 0.1508,
      "step": 1238,
      "task_loss": 0.07494504749774933
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7379264527755719,
      "compression_loss": 0.0,
      "distillation_loss": 0.216340109705925,
      "epoch": 1.18,
      "learning_rate": 4.661056737723349e-05,
      "loss": 0.2053,
      "step": 1239,
      "task_loss": 0.10613324493169785
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7379523635365667,
      "compression_loss": 0.0,
      "distillation_loss": 0.4159829914569855,
      "epoch": 1.18,
      "learning_rate": 4.660520832189732e-05,
      "loss": 0.4001,
      "step": 1240,
      "task_loss": 0.257106214761734
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7379782670861056,
      "compression_loss": 0.0,
      "distillation_loss": 0.04487886279821396,
      "epoch": 1.18,
      "learning_rate": 4.6599845341871005e-05,
      "loss": 0.0454,
      "step": 1241,
      "task_loss": 0.050378091633319855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7380041634251924,
      "compression_loss": 0.0,
      "distillation_loss": 0.13262313604354858,
      "epoch": 1.18,
      "learning_rate": 4.6594478438128757e-05,
      "loss": 0.1244,
      "step": 1242,
      "task_loss": 0.05053956061601639
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7380300525548307,
      "compression_loss": 0.0,
      "distillation_loss": 0.06497453153133392,
      "epoch": 1.18,
      "learning_rate": 4.6589107611645497e-05,
      "loss": 0.0752,
      "step": 1243,
      "task_loss": 0.16748173534870148
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7380559344760241,
      "compression_loss": 0.0,
      "distillation_loss": 0.16354522109031677,
      "epoch": 1.18,
      "learning_rate": 4.658373286339688e-05,
      "loss": 0.1567,
      "step": 1244,
      "task_loss": 0.09554801136255264
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7380818091897765,
      "compression_loss": 0.0,
      "distillation_loss": 0.11697202920913696,
      "epoch": 1.18,
      "learning_rate": 4.6578354194359227e-05,
      "loss": 0.1177,
      "step": 1245,
      "task_loss": 0.12439711391925812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7381076766970915,
      "compression_loss": 0.0,
      "distillation_loss": 0.139469712972641,
      "epoch": 1.18,
      "learning_rate": 4.657297160550961e-05,
      "loss": 0.1305,
      "step": 1246,
      "task_loss": 0.04955626651644707
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7381335369989727,
      "compression_loss": 0.0,
      "distillation_loss": 0.1701054573059082,
      "epoch": 1.18,
      "learning_rate": 4.656758509782582e-05,
      "loss": 0.1753,
      "step": 1247,
      "task_loss": 0.22217293083667755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7381593900964238,
      "compression_loss": 0.0,
      "distillation_loss": 0.22485214471817017,
      "epoch": 1.19,
      "learning_rate": 4.6562194672286306e-05,
      "loss": 0.2262,
      "step": 1248,
      "task_loss": 0.23852680623531342
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7381852359904486,
      "compression_loss": 0.0,
      "distillation_loss": 0.06678508222103119,
      "epoch": 1.19,
      "learning_rate": 4.65568003298703e-05,
      "loss": 0.0617,
      "step": 1249,
      "task_loss": 0.016354495659470558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7382110746820507,
      "compression_loss": 0.0,
      "distillation_loss": 0.11139645427465439,
      "epoch": 1.19,
      "learning_rate": 4.655140207155769e-05,
      "loss": 0.1023,
      "step": 1250,
      "task_loss": 0.020297054201364517
    },
    {
      "epoch": 1.19,
      "eval_accuracy": 0.8979357798165137,
      "eval_loss": 0.4085277318954468,
      "eval_runtime": 18.1765,
      "eval_samples_per_second": 47.974,
      "eval_steps_per_second": 5.997,
      "step": 1250
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7382369061722337,
      "compression_loss": 0.0,
      "distillation_loss": 0.11869402974843979,
      "epoch": 1.19,
      "learning_rate": 4.65459998983291e-05,
      "loss": 0.1219,
      "step": 1251,
      "task_loss": 0.15096434950828552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7382627304620014,
      "compression_loss": 0.0,
      "distillation_loss": 0.22661426663398743,
      "epoch": 1.19,
      "learning_rate": 4.6540593811165866e-05,
      "loss": 0.2328,
      "step": 1252,
      "task_loss": 0.28801366686820984
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7382885475523575,
      "compression_loss": 0.0,
      "distillation_loss": 0.03438243269920349,
      "epoch": 1.19,
      "learning_rate": 4.653518381105002e-05,
      "loss": 0.0423,
      "step": 1253,
      "task_loss": 0.11402395367622375
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7383143574443055,
      "compression_loss": 0.0,
      "distillation_loss": 0.03891141712665558,
      "epoch": 1.19,
      "learning_rate": 4.6529769898964325e-05,
      "loss": 0.0358,
      "step": 1254,
      "task_loss": 0.007961155846714973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7383401601388494,
      "compression_loss": 0.0,
      "distillation_loss": 0.036808304488658905,
      "epoch": 1.19,
      "learning_rate": 4.652435207589224e-05,
      "loss": 0.0468,
      "step": 1255,
      "task_loss": 0.1369296759366989
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7383659556369926,
      "compression_loss": 0.0,
      "distillation_loss": 0.07549430429935455,
      "epoch": 1.19,
      "learning_rate": 4.651893034281793e-05,
      "loss": 0.0702,
      "step": 1256,
      "task_loss": 0.02279655635356903
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7383917439397388,
      "compression_loss": 0.0,
      "distillation_loss": 0.0869937464594841,
      "epoch": 1.19,
      "learning_rate": 4.6513504700726293e-05,
      "loss": 0.0966,
      "step": 1257,
      "task_loss": 0.1825573891401291
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7384175250480919,
      "compression_loss": 0.0,
      "distillation_loss": 0.1446138620376587,
      "epoch": 1.19,
      "learning_rate": 4.650807515060291e-05,
      "loss": 0.1397,
      "step": 1258,
      "task_loss": 0.09526462852954865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7384432989630553,
      "compression_loss": 0.0,
      "distillation_loss": 0.15781186521053314,
      "epoch": 1.2,
      "learning_rate": 4.650264169343411e-05,
      "loss": 0.1688,
      "step": 1259,
      "task_loss": 0.26753732562065125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7384690656856329,
      "compression_loss": 0.0,
      "distillation_loss": 0.06009237468242645,
      "epoch": 1.2,
      "learning_rate": 4.6497204330206874e-05,
      "loss": 0.0683,
      "step": 1260,
      "task_loss": 0.14257624745368958
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7384948252168283,
      "compression_loss": 0.0,
      "distillation_loss": 0.1592351347208023,
      "epoch": 1.2,
      "learning_rate": 4.649176306190895e-05,
      "loss": 0.1589,
      "step": 1261,
      "task_loss": 0.15542340278625488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7385205775576451,
      "compression_loss": 0.0,
      "distillation_loss": 0.3260449469089508,
      "epoch": 1.2,
      "learning_rate": 4.648631788952874e-05,
      "loss": 0.3149,
      "step": 1262,
      "task_loss": 0.21453765034675598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7385463227090872,
      "compression_loss": 0.0,
      "distillation_loss": 0.03898075968027115,
      "epoch": 1.2,
      "learning_rate": 4.6480868814055424e-05,
      "loss": 0.0397,
      "step": 1263,
      "task_loss": 0.04632849618792534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7385720606721581,
      "compression_loss": 0.0,
      "distillation_loss": 0.3080974519252777,
      "epoch": 1.2,
      "learning_rate": 4.647541583647883e-05,
      "loss": 0.2963,
      "step": 1264,
      "task_loss": 0.1906224638223648
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7385977914478615,
      "compression_loss": 0.0,
      "distillation_loss": 0.1396906077861786,
      "epoch": 1.2,
      "learning_rate": 4.646995895778952e-05,
      "loss": 0.1367,
      "step": 1265,
      "task_loss": 0.11023291200399399
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7386235150372011,
      "compression_loss": 0.0,
      "distillation_loss": 0.16227009892463684,
      "epoch": 1.2,
      "learning_rate": 4.646449817897876e-05,
      "loss": 0.1593,
      "step": 1266,
      "task_loss": 0.13288387656211853
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7386492314411807,
      "compression_loss": 0.0,
      "distillation_loss": 0.18255099654197693,
      "epoch": 1.2,
      "learning_rate": 4.645903350103855e-05,
      "loss": 0.1828,
      "step": 1267,
      "task_loss": 0.18547126650810242
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7386749406608037,
      "compression_loss": 0.0,
      "distillation_loss": 0.13566656410694122,
      "epoch": 1.2,
      "learning_rate": 4.6453564924961544e-05,
      "loss": 0.1234,
      "step": 1268,
      "task_loss": 0.012727364897727966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7387006426970741,
      "compression_loss": 0.0,
      "distillation_loss": 0.10819855332374573,
      "epoch": 1.21,
      "learning_rate": 4.644809245174114e-05,
      "loss": 0.1009,
      "step": 1269,
      "task_loss": 0.034790411591529846
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7387263375509955,
      "compression_loss": 0.0,
      "distillation_loss": 0.2560357451438904,
      "epoch": 1.21,
      "learning_rate": 4.6442616082371466e-05,
      "loss": 0.2434,
      "step": 1270,
      "task_loss": 0.12986107170581818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7387520252235714,
      "compression_loss": 0.0,
      "distillation_loss": 0.06887871772050858,
      "epoch": 1.21,
      "learning_rate": 4.64371358178473e-05,
      "loss": 0.0706,
      "step": 1271,
      "task_loss": 0.08568105101585388
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7387777057158056,
      "compression_loss": 0.0,
      "distillation_loss": 0.20982679724693298,
      "epoch": 1.21,
      "learning_rate": 4.6431651659164174e-05,
      "loss": 0.2139,
      "step": 1272,
      "task_loss": 0.25017955899238586
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7388033790287019,
      "compression_loss": 0.0,
      "distillation_loss": 0.2947445511817932,
      "epoch": 1.21,
      "learning_rate": 4.6426163607318305e-05,
      "loss": 0.2797,
      "step": 1273,
      "task_loss": 0.14412230253219604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7388290451632638,
      "compression_loss": 0.0,
      "distillation_loss": 0.10616719722747803,
      "epoch": 1.21,
      "learning_rate": 4.642067166330663e-05,
      "loss": 0.1096,
      "step": 1274,
      "task_loss": 0.1407913863658905
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.738854704120495,
      "compression_loss": 0.0,
      "distillation_loss": 0.262935608625412,
      "epoch": 1.21,
      "learning_rate": 4.6415175828126786e-05,
      "loss": 0.2522,
      "step": 1275,
      "task_loss": 0.15595991909503937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7388803559013993,
      "compression_loss": 0.0,
      "distillation_loss": 0.08681028336286545,
      "epoch": 1.21,
      "learning_rate": 4.640967610277711e-05,
      "loss": 0.0923,
      "step": 1276,
      "task_loss": 0.14123068749904633
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7389060005069803,
      "compression_loss": 0.0,
      "distillation_loss": 0.2555930018424988,
      "epoch": 1.21,
      "learning_rate": 4.640417248825667e-05,
      "loss": 0.2431,
      "step": 1277,
      "task_loss": 0.13036002218723297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7389316379382417,
      "compression_loss": 0.0,
      "distillation_loss": 0.03256315737962723,
      "epoch": 1.21,
      "learning_rate": 4.63986649855652e-05,
      "loss": 0.0301,
      "step": 1278,
      "task_loss": 0.007671518251299858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7389572681961872,
      "compression_loss": 0.0,
      "distillation_loss": 0.06931599974632263,
      "epoch": 1.21,
      "learning_rate": 4.639315359570319e-05,
      "loss": 0.0782,
      "step": 1279,
      "task_loss": 0.1577843278646469
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7389828912818205,
      "compression_loss": 0.0,
      "distillation_loss": 0.09818733483552933,
      "epoch": 1.22,
      "learning_rate": 4.6387638319671786e-05,
      "loss": 0.1179,
      "step": 1280,
      "task_loss": 0.2952803671360016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7390085071961452,
      "compression_loss": 0.0,
      "distillation_loss": 0.055818259716033936,
      "epoch": 1.22,
      "learning_rate": 4.6382119158472895e-05,
      "loss": 0.0511,
      "step": 1281,
      "task_loss": 0.009016238152980804
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.739034115940165,
      "compression_loss": 0.0,
      "distillation_loss": 0.283183753490448,
      "epoch": 1.22,
      "learning_rate": 4.637659611310907e-05,
      "loss": 0.2704,
      "step": 1282,
      "task_loss": 0.15582668781280518
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7390597175148836,
      "compression_loss": 0.0,
      "distillation_loss": 0.08733191341161728,
      "epoch": 1.22,
      "learning_rate": 4.637106918458361e-05,
      "loss": 0.0902,
      "step": 1283,
      "task_loss": 0.11557549238204956
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7390853119213048,
      "compression_loss": 0.0,
      "distillation_loss": 0.05915270373225212,
      "epoch": 1.22,
      "learning_rate": 4.636553837390051e-05,
      "loss": 0.0552,
      "step": 1284,
      "task_loss": 0.01949235051870346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.739110899160432,
      "compression_loss": 0.0,
      "distillation_loss": 0.22277674078941345,
      "epoch": 1.22,
      "learning_rate": 4.636000368206447e-05,
      "loss": 0.2188,
      "step": 1285,
      "task_loss": 0.18264567852020264
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7391364792332692,
      "compression_loss": 0.0,
      "distillation_loss": 0.07782046496868134,
      "epoch": 1.22,
      "learning_rate": 4.6354465110080885e-05,
      "loss": 0.0838,
      "step": 1286,
      "task_loss": 0.1375577300786972
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7391620521408199,
      "compression_loss": 0.0,
      "distillation_loss": 0.1343613713979721,
      "epoch": 1.22,
      "learning_rate": 4.6348922658955874e-05,
      "loss": 0.1423,
      "step": 1287,
      "task_loss": 0.2140950858592987
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7391876178840878,
      "compression_loss": 0.0,
      "distillation_loss": 0.24139809608459473,
      "epoch": 1.22,
      "learning_rate": 4.634337632969624e-05,
      "loss": 0.2376,
      "step": 1288,
      "task_loss": 0.20292049646377563
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7392131764640766,
      "compression_loss": 0.0,
      "distillation_loss": 0.1293402463197708,
      "epoch": 1.22,
      "learning_rate": 4.6337826123309505e-05,
      "loss": 0.123,
      "step": 1289,
      "task_loss": 0.06586703658103943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.73923872788179,
      "compression_loss": 0.0,
      "distillation_loss": 0.09735876321792603,
      "epoch": 1.23,
      "learning_rate": 4.6332272040803895e-05,
      "loss": 0.1097,
      "step": 1290,
      "task_loss": 0.22072833776474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7392642721382316,
      "compression_loss": 0.0,
      "distillation_loss": 0.30776453018188477,
      "epoch": 1.23,
      "learning_rate": 4.632671408318833e-05,
      "loss": 0.2998,
      "step": 1291,
      "task_loss": 0.22781196236610413
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7392898092344052,
      "compression_loss": 0.0,
      "distillation_loss": 0.13507933914661407,
      "epoch": 1.23,
      "learning_rate": 4.6321152251472435e-05,
      "loss": 0.1333,
      "step": 1292,
      "task_loss": 0.11699161678552628
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7393153391713145,
      "compression_loss": 0.0,
      "distillation_loss": 0.30968376994132996,
      "epoch": 1.23,
      "learning_rate": 4.6315586546666556e-05,
      "loss": 0.3044,
      "step": 1293,
      "task_loss": 0.2563600540161133
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7393408619499631,
      "compression_loss": 0.0,
      "distillation_loss": 0.15666070580482483,
      "epoch": 1.23,
      "learning_rate": 4.631001696978172e-05,
      "loss": 0.1524,
      "step": 1294,
      "task_loss": 0.11410398036241531
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7393663775713546,
      "compression_loss": 0.0,
      "distillation_loss": 0.21059119701385498,
      "epoch": 1.23,
      "learning_rate": 4.630444352182968e-05,
      "loss": 0.1988,
      "step": 1295,
      "task_loss": 0.09313502162694931
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7393918860364929,
      "compression_loss": 0.0,
      "distillation_loss": 0.2058752477169037,
      "epoch": 1.23,
      "learning_rate": 4.6298866203822865e-05,
      "loss": 0.1976,
      "step": 1296,
      "task_loss": 0.12271516025066376
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7394173873463814,
      "compression_loss": 0.0,
      "distillation_loss": 0.03244621679186821,
      "epoch": 1.23,
      "learning_rate": 4.629328501677442e-05,
      "loss": 0.0299,
      "step": 1297,
      "task_loss": 0.007005665451288223
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7394428815020241,
      "compression_loss": 0.0,
      "distillation_loss": 0.05955211818218231,
      "epoch": 1.23,
      "learning_rate": 4.6287699961698214e-05,
      "loss": 0.0587,
      "step": 1298,
      "task_loss": 0.050543755292892456
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7394683685044244,
      "compression_loss": 0.0,
      "distillation_loss": 0.10339696705341339,
      "epoch": 1.23,
      "learning_rate": 4.6282111039608784e-05,
      "loss": 0.1048,
      "step": 1299,
      "task_loss": 0.1170087605714798
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7394938483545862,
      "compression_loss": 0.0,
      "distillation_loss": 0.13691368699073792,
      "epoch": 1.23,
      "learning_rate": 4.6276518251521384e-05,
      "loss": 0.148,
      "step": 1300,
      "task_loss": 0.24737101793289185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.739519321053513,
      "compression_loss": 0.0,
      "distillation_loss": 0.2226058393716812,
      "epoch": 1.24,
      "learning_rate": 4.6270921598451974e-05,
      "loss": 0.2199,
      "step": 1301,
      "task_loss": 0.19590626657009125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7395447866022086,
      "compression_loss": 0.0,
      "distillation_loss": 0.25497111678123474,
      "epoch": 1.24,
      "learning_rate": 4.6265321081417223e-05,
      "loss": 0.2433,
      "step": 1302,
      "task_loss": 0.13829627633094788
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7395702450016767,
      "compression_loss": 0.0,
      "distillation_loss": 0.20901359617710114,
      "epoch": 1.24,
      "learning_rate": 4.625971670143447e-05,
      "loss": 0.2123,
      "step": 1303,
      "task_loss": 0.242196723818779
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7395956962529209,
      "compression_loss": 0.0,
      "distillation_loss": 0.08364978432655334,
      "epoch": 1.24,
      "learning_rate": 4.625410845952181e-05,
      "loss": 0.0823,
      "step": 1304,
      "task_loss": 0.07054957747459412
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7396211403569448,
      "compression_loss": 0.0,
      "distillation_loss": 0.18630865216255188,
      "epoch": 1.24,
      "learning_rate": 4.6248496356697966e-05,
      "loss": 0.1769,
      "step": 1305,
      "task_loss": 0.09224953502416611
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7396465773147524,
      "compression_loss": 0.0,
      "distillation_loss": 0.2079707384109497,
      "epoch": 1.24,
      "learning_rate": 4.6242880393982436e-05,
      "loss": 0.1995,
      "step": 1306,
      "task_loss": 0.12356877326965332
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.739672007127347,
      "compression_loss": 0.0,
      "distillation_loss": 0.2150033861398697,
      "epoch": 1.24,
      "learning_rate": 4.623726057239537e-05,
      "loss": 0.2303,
      "step": 1307,
      "task_loss": 0.3683556020259857
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7396974297957326,
      "compression_loss": 0.0,
      "distillation_loss": 0.21252375841140747,
      "epoch": 1.24,
      "learning_rate": 4.623163689295764e-05,
      "loss": 0.2039,
      "step": 1308,
      "task_loss": 0.12652674317359924
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7397228453209127,
      "compression_loss": 0.0,
      "distillation_loss": 0.21421362459659576,
      "epoch": 1.24,
      "learning_rate": 4.6226009356690825e-05,
      "loss": 0.2026,
      "step": 1309,
      "task_loss": 0.09814447164535522
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7397482537038911,
      "compression_loss": 0.0,
      "distillation_loss": 0.20858995616436005,
      "epoch": 1.24,
      "learning_rate": 4.622037796461718e-05,
      "loss": 0.198,
      "step": 1310,
      "task_loss": 0.10299951583147049
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7397736549456713,
      "compression_loss": 0.0,
      "distillation_loss": 0.2604760527610779,
      "epoch": 1.25,
      "learning_rate": 4.621474271775968e-05,
      "loss": 0.2562,
      "step": 1311,
      "task_loss": 0.21733561158180237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.739799049047257,
      "compression_loss": 0.0,
      "distillation_loss": 0.07183189690113068,
      "epoch": 1.25,
      "learning_rate": 4.620910361714199e-05,
      "loss": 0.068,
      "step": 1312,
      "task_loss": 0.03307514637708664
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7398244360096521,
      "compression_loss": 0.0,
      "distillation_loss": 0.19191870093345642,
      "epoch": 1.25,
      "learning_rate": 4.620346066378849e-05,
      "loss": 0.1822,
      "step": 1313,
      "task_loss": 0.09521719813346863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7398498158338602,
      "compression_loss": 0.0,
      "distillation_loss": 0.1862865388393402,
      "epoch": 1.25,
      "learning_rate": 4.619781385872424e-05,
      "loss": 0.1838,
      "step": 1314,
      "task_loss": 0.16108401119709015
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7398751885208849,
      "compression_loss": 0.0,
      "distillation_loss": 0.28807079792022705,
      "epoch": 1.25,
      "learning_rate": 4.6192163202975013e-05,
      "loss": 0.2742,
      "step": 1315,
      "task_loss": 0.14911451935768127
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7399005540717298,
      "compression_loss": 0.0,
      "distillation_loss": 0.07601656764745712,
      "epoch": 1.25,
      "learning_rate": 4.618650869756728e-05,
      "loss": 0.071,
      "step": 1316,
      "task_loss": 0.025538241490721703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7399259124873989,
      "compression_loss": 0.0,
      "distillation_loss": 0.08642975986003876,
      "epoch": 1.25,
      "learning_rate": 4.6180850343528205e-05,
      "loss": 0.0799,
      "step": 1317,
      "task_loss": 0.021204736083745956
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7399512637688955,
      "compression_loss": 0.0,
      "distillation_loss": 0.10312186181545258,
      "epoch": 1.25,
      "learning_rate": 4.617518814188565e-05,
      "loss": 0.0966,
      "step": 1318,
      "task_loss": 0.037991687655448914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7399766079172236,
      "compression_loss": 0.0,
      "distillation_loss": 0.2704732418060303,
      "epoch": 1.25,
      "learning_rate": 4.6169522093668196e-05,
      "loss": 0.2797,
      "step": 1319,
      "task_loss": 0.3626573085784912
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7400019449333867,
      "compression_loss": 0.0,
      "distillation_loss": 0.10321257263422012,
      "epoch": 1.25,
      "learning_rate": 4.61638521999051e-05,
      "loss": 0.0959,
      "step": 1320,
      "task_loss": 0.030156267806887627
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7400272748183885,
      "compression_loss": 0.0,
      "distillation_loss": 0.5204591751098633,
      "epoch": 1.25,
      "learning_rate": 4.6158178461626323e-05,
      "loss": 0.5081,
      "step": 1321,
      "task_loss": 0.3965243101119995
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7400525975732327,
      "compression_loss": 0.0,
      "distillation_loss": 0.23917892575263977,
      "epoch": 1.26,
      "learning_rate": 4.615250087986254e-05,
      "loss": 0.2394,
      "step": 1322,
      "task_loss": 0.24161620438098907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.740077913198923,
      "compression_loss": 0.0,
      "distillation_loss": 0.09203709661960602,
      "epoch": 1.26,
      "learning_rate": 4.6146819455645086e-05,
      "loss": 0.0865,
      "step": 1323,
      "task_loss": 0.037142425775527954
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7401032216964631,
      "compression_loss": 0.0,
      "distillation_loss": 0.10284596681594849,
      "epoch": 1.26,
      "learning_rate": 4.614113419000604e-05,
      "loss": 0.1036,
      "step": 1324,
      "task_loss": 0.11075641214847565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7401285230668567,
      "compression_loss": 0.0,
      "distillation_loss": 0.2411002814769745,
      "epoch": 1.26,
      "learning_rate": 4.613544508397815e-05,
      "loss": 0.2294,
      "step": 1325,
      "task_loss": 0.12409268319606781
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7401538173111074,
      "compression_loss": 0.0,
      "distillation_loss": 0.08729246258735657,
      "epoch": 1.26,
      "learning_rate": 4.6129752138594874e-05,
      "loss": 0.0873,
      "step": 1326,
      "task_loss": 0.0873517096042633
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.740179104430219,
      "compression_loss": 0.0,
      "distillation_loss": 0.13888351619243622,
      "epoch": 1.26,
      "learning_rate": 4.612405535489036e-05,
      "loss": 0.1341,
      "step": 1327,
      "task_loss": 0.09130087494850159
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.740204384425195,
      "compression_loss": 0.0,
      "distillation_loss": 0.1486589014530182,
      "epoch": 1.26,
      "learning_rate": 4.611835473389945e-05,
      "loss": 0.148,
      "step": 1328,
      "task_loss": 0.14163073897361755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7402296572970392,
      "compression_loss": 0.0,
      "distillation_loss": 0.352372944355011,
      "epoch": 1.26,
      "learning_rate": 4.61126502766577e-05,
      "loss": 0.3364,
      "step": 1329,
      "task_loss": 0.192458838224411
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7402549230467552,
      "compression_loss": 0.0,
      "distillation_loss": 0.09347756952047348,
      "epoch": 1.26,
      "learning_rate": 4.6106941984201344e-05,
      "loss": 0.1093,
      "step": 1330,
      "task_loss": 0.251775860786438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7402801816753469,
      "compression_loss": 0.0,
      "distillation_loss": 0.16438814997673035,
      "epoch": 1.26,
      "learning_rate": 4.610122985756733e-05,
      "loss": 0.158,
      "step": 1331,
      "task_loss": 0.10030940920114517
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7403054331838178,
      "compression_loss": 0.0,
      "distillation_loss": 0.186666801571846,
      "epoch": 1.26,
      "learning_rate": 4.609551389779328e-05,
      "loss": 0.1849,
      "step": 1332,
      "task_loss": 0.16852930188179016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7403306775731715,
      "compression_loss": 0.0,
      "distillation_loss": 0.0614393912255764,
      "epoch": 1.27,
      "learning_rate": 4.6089794105917544e-05,
      "loss": 0.063,
      "step": 1333,
      "task_loss": 0.0771103948354721
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7403559148444119,
      "compression_loss": 0.0,
      "distillation_loss": 0.28277117013931274,
      "epoch": 1.27,
      "learning_rate": 4.6084070482979135e-05,
      "loss": 0.2787,
      "step": 1334,
      "task_loss": 0.24248361587524414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7403811449985426,
      "compression_loss": 0.0,
      "distillation_loss": 0.24766165018081665,
      "epoch": 1.27,
      "learning_rate": 4.607834303001778e-05,
      "loss": 0.2406,
      "step": 1335,
      "task_loss": 0.17734551429748535
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7404063680365671,
      "compression_loss": 0.0,
      "distillation_loss": 0.21882733702659607,
      "epoch": 1.27,
      "learning_rate": 4.60726117480739e-05,
      "loss": 0.2158,
      "step": 1336,
      "task_loss": 0.18877999484539032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7404315839594894,
      "compression_loss": 0.0,
      "distillation_loss": 0.1295139193534851,
      "epoch": 1.27,
      "learning_rate": 4.6066876638188604e-05,
      "loss": 0.1317,
      "step": 1337,
      "task_loss": 0.151097372174263
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.740456792768313,
      "compression_loss": 0.0,
      "distillation_loss": 0.199735626578331,
      "epoch": 1.27,
      "learning_rate": 4.606113770140371e-05,
      "loss": 0.1898,
      "step": 1338,
      "task_loss": 0.10075341165065765
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7404819944640416,
      "compression_loss": 0.0,
      "distillation_loss": 0.2485513985157013,
      "epoch": 1.27,
      "learning_rate": 4.605539493876173e-05,
      "loss": 0.2438,
      "step": 1339,
      "task_loss": 0.20112858712673187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7405071890476789,
      "compression_loss": 0.0,
      "distillation_loss": 0.2236090898513794,
      "epoch": 1.27,
      "learning_rate": 4.604964835130585e-05,
      "loss": 0.2099,
      "step": 1340,
      "task_loss": 0.08700668066740036
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7405323765202285,
      "compression_loss": 0.0,
      "distillation_loss": 0.062271565198898315,
      "epoch": 1.27,
      "learning_rate": 4.6043897940079964e-05,
      "loss": 0.0583,
      "step": 1341,
      "task_loss": 0.022575678303837776
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7405575568826942,
      "compression_loss": 0.0,
      "distillation_loss": 0.08404142409563065,
      "epoch": 1.27,
      "learning_rate": 4.603814370612867e-05,
      "loss": 0.078,
      "step": 1342,
      "task_loss": 0.02370348386466503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7405827301360796,
      "compression_loss": 0.0,
      "distillation_loss": 0.2024337202310562,
      "epoch": 1.28,
      "learning_rate": 4.603238565049726e-05,
      "loss": 0.1899,
      "step": 1343,
      "task_loss": 0.07661712914705276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7406078962813885,
      "compression_loss": 0.0,
      "distillation_loss": 0.29442059993743896,
      "epoch": 1.28,
      "learning_rate": 4.60266237742317e-05,
      "loss": 0.279,
      "step": 1344,
      "task_loss": 0.1398637890815735
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7406330553196244,
      "compression_loss": 0.0,
      "distillation_loss": 0.03961509093642235,
      "epoch": 1.28,
      "learning_rate": 4.602085807837866e-05,
      "loss": 0.0527,
      "step": 1345,
      "task_loss": 0.1707422137260437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7406582072517911,
      "compression_loss": 0.0,
      "distillation_loss": 0.3172229826450348,
      "epoch": 1.28,
      "learning_rate": 4.601508856398552e-05,
      "loss": 0.3057,
      "step": 1346,
      "task_loss": 0.20200252532958984
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7406833520788924,
      "compression_loss": 0.0,
      "distillation_loss": 0.1311604082584381,
      "epoch": 1.28,
      "learning_rate": 4.6009315232100324e-05,
      "loss": 0.1316,
      "step": 1347,
      "task_loss": 0.13595682382583618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7407084898019316,
      "compression_loss": 0.0,
      "distillation_loss": 0.08850497007369995,
      "epoch": 1.28,
      "learning_rate": 4.600353808377184e-05,
      "loss": 0.0874,
      "step": 1348,
      "task_loss": 0.07742930948734283
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7407336204219128,
      "compression_loss": 0.0,
      "distillation_loss": 0.03612298145890236,
      "epoch": 1.28,
      "learning_rate": 4.599775712004951e-05,
      "loss": 0.0333,
      "step": 1349,
      "task_loss": 0.007720911875367165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7407587439398395,
      "compression_loss": 0.0,
      "distillation_loss": 0.552012026309967,
      "epoch": 1.28,
      "learning_rate": 4.599197234198347e-05,
      "loss": 0.5405,
      "step": 1350,
      "task_loss": 0.4371148347854614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7407838603567154,
      "compression_loss": 0.0,
      "distillation_loss": 0.07328462600708008,
      "epoch": 1.28,
      "learning_rate": 4.5986183750624555e-05,
      "loss": 0.0754,
      "step": 1351,
      "task_loss": 0.09475519508123398
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7408089696735441,
      "compression_loss": 0.0,
      "distillation_loss": 0.2518464922904968,
      "epoch": 1.28,
      "learning_rate": 4.5980391347024296e-05,
      "loss": 0.2416,
      "step": 1352,
      "task_loss": 0.14951427280902863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7408340718913294,
      "compression_loss": 0.0,
      "distillation_loss": 0.28138667345046997,
      "epoch": 1.28,
      "learning_rate": 4.59745951322349e-05,
      "loss": 0.2688,
      "step": 1353,
      "task_loss": 0.15557241439819336
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7408591670110749,
      "compression_loss": 0.0,
      "distillation_loss": 0.11055044829845428,
      "epoch": 1.29,
      "learning_rate": 4.596879510730929e-05,
      "loss": 0.1117,
      "step": 1354,
      "task_loss": 0.12174604833126068
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7408842550337844,
      "compression_loss": 0.0,
      "distillation_loss": 0.0641191229224205,
      "epoch": 1.29,
      "learning_rate": 4.596299127330106e-05,
      "loss": 0.0583,
      "step": 1355,
      "task_loss": 0.0056402478367090225
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7409093359604615,
      "compression_loss": 0.0,
      "distillation_loss": 0.06709885597229004,
      "epoch": 1.29,
      "learning_rate": 4.59571836312645e-05,
      "loss": 0.0675,
      "step": 1356,
      "task_loss": 0.07135514914989471
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7409344097921098,
      "compression_loss": 0.0,
      "distillation_loss": 0.1536739617586136,
      "epoch": 1.29,
      "learning_rate": 4.595137218225461e-05,
      "loss": 0.1479,
      "step": 1357,
      "task_loss": 0.09615078568458557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7409594765297332,
      "compression_loss": 0.0,
      "distillation_loss": 0.2070506066083908,
      "epoch": 1.29,
      "learning_rate": 4.594555692732706e-05,
      "loss": 0.2038,
      "step": 1358,
      "task_loss": 0.1743832379579544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.740984536174335,
      "compression_loss": 0.0,
      "distillation_loss": 0.24169021844863892,
      "epoch": 1.29,
      "learning_rate": 4.593973786753821e-05,
      "loss": 0.2442,
      "step": 1359,
      "task_loss": 0.2664361894130707
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7410095887269194,
      "compression_loss": 0.0,
      "distillation_loss": 0.2801350951194763,
      "epoch": 1.29,
      "learning_rate": 4.593391500394514e-05,
      "loss": 0.2794,
      "step": 1360,
      "task_loss": 0.2726552486419678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7410346341884897,
      "compression_loss": 0.0,
      "distillation_loss": 0.09238822013139725,
      "epoch": 1.29,
      "learning_rate": 4.5928088337605586e-05,
      "loss": 0.0886,
      "step": 1361,
      "task_loss": 0.05469810217618942
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7410596725600497,
      "compression_loss": 0.0,
      "distillation_loss": 0.09862814098596573,
      "epoch": 1.29,
      "learning_rate": 4.5922257869578e-05,
      "loss": 0.0927,
      "step": 1362,
      "task_loss": 0.0397910512983799
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.741084703842603,
      "compression_loss": 0.0,
      "distillation_loss": 0.18891645967960358,
      "epoch": 1.29,
      "learning_rate": 4.5916423600921496e-05,
      "loss": 0.1851,
      "step": 1363,
      "task_loss": 0.1504015326499939
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7411097280371535,
      "compression_loss": 0.0,
      "distillation_loss": 0.34903091192245483,
      "epoch": 1.3,
      "learning_rate": 4.591058553269593e-05,
      "loss": 0.3357,
      "step": 1364,
      "task_loss": 0.21621909737586975
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7411347451447046,
      "compression_loss": 0.0,
      "distillation_loss": 0.0994444340467453,
      "epoch": 1.3,
      "learning_rate": 4.590474366596178e-05,
      "loss": 0.1141,
      "step": 1365,
      "task_loss": 0.24601247906684875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7411597551662602,
      "compression_loss": 0.0,
      "distillation_loss": 0.11619491875171661,
      "epoch": 1.3,
      "learning_rate": 4.589889800178026e-05,
      "loss": 0.1112,
      "step": 1366,
      "task_loss": 0.06617462635040283
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7411847581028239,
      "compression_loss": 0.0,
      "distillation_loss": 0.08728669583797455,
      "epoch": 1.3,
      "learning_rate": 4.589304854121329e-05,
      "loss": 0.0875,
      "step": 1367,
      "task_loss": 0.08968639373779297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7412097539553995,
      "compression_loss": 0.0,
      "distillation_loss": 0.13692383468151093,
      "epoch": 1.3,
      "learning_rate": 4.588719528532342e-05,
      "loss": 0.1391,
      "step": 1368,
      "task_loss": 0.15917925536632538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7412347427249903,
      "compression_loss": 0.0,
      "distillation_loss": 0.2873116433620453,
      "epoch": 1.3,
      "learning_rate": 4.588133823517392e-05,
      "loss": 0.2825,
      "step": 1369,
      "task_loss": 0.23926284909248352
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7412597244126005,
      "compression_loss": 0.0,
      "distillation_loss": 0.22201889753341675,
      "epoch": 1.3,
      "learning_rate": 4.587547739182878e-05,
      "loss": 0.2193,
      "step": 1370,
      "task_loss": 0.19527406990528107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7412846990192334,
      "compression_loss": 0.0,
      "distillation_loss": 0.15326130390167236,
      "epoch": 1.3,
      "learning_rate": 4.586961275635263e-05,
      "loss": 0.1554,
      "step": 1371,
      "task_loss": 0.17474180459976196
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7413096665458929,
      "compression_loss": 0.0,
      "distillation_loss": 0.1608862280845642,
      "epoch": 1.3,
      "learning_rate": 4.586374432981081e-05,
      "loss": 0.1622,
      "step": 1372,
      "task_loss": 0.17383195459842682
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7413346269935825,
      "compression_loss": 0.0,
      "distillation_loss": 0.05019424855709076,
      "epoch": 1.3,
      "learning_rate": 4.585787211326935e-05,
      "loss": 0.0561,
      "step": 1373,
      "task_loss": 0.1092485561966896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7413595803633061,
      "compression_loss": 0.0,
      "distillation_loss": 0.45158857107162476,
      "epoch": 1.3,
      "learning_rate": 4.5851996107794975e-05,
      "loss": 0.4364,
      "step": 1374,
      "task_loss": 0.2999117970466614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7413845266560671,
      "compression_loss": 0.0,
      "distillation_loss": 0.128067284822464,
      "epoch": 1.31,
      "learning_rate": 4.584611631445508e-05,
      "loss": 0.1328,
      "step": 1375,
      "task_loss": 0.1756190061569214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7414094658728695,
      "compression_loss": 0.0,
      "distillation_loss": 0.1457686424255371,
      "epoch": 1.31,
      "learning_rate": 4.5840232734317754e-05,
      "loss": 0.1414,
      "step": 1376,
      "task_loss": 0.10216192901134491
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7414343980147168,
      "compression_loss": 0.0,
      "distillation_loss": 0.33278051018714905,
      "epoch": 1.31,
      "learning_rate": 4.583434536845179e-05,
      "loss": 0.3246,
      "step": 1377,
      "task_loss": 0.2506526708602905
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7414593230826126,
      "compression_loss": 0.0,
      "distillation_loss": 0.1251896768808365,
      "epoch": 1.31,
      "learning_rate": 4.5828454217926654e-05,
      "loss": 0.1325,
      "step": 1378,
      "task_loss": 0.19821739196777344
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7414842410775608,
      "compression_loss": 0.0,
      "distillation_loss": 0.18593353033065796,
      "epoch": 1.31,
      "learning_rate": 4.5822559283812496e-05,
      "loss": 0.1785,
      "step": 1379,
      "task_loss": 0.11133649945259094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.741509152000565,
      "compression_loss": 0.0,
      "distillation_loss": 0.058180175721645355,
      "epoch": 1.31,
      "learning_rate": 4.581666056718016e-05,
      "loss": 0.0539,
      "step": 1380,
      "task_loss": 0.015172762796282768
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7415340558526288,
      "compression_loss": 0.0,
      "distillation_loss": 0.2561149001121521,
      "epoch": 1.31,
      "learning_rate": 4.5810758069101175e-05,
      "loss": 0.2525,
      "step": 1381,
      "task_loss": 0.2195451259613037
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7415589526347559,
      "compression_loss": 0.0,
      "distillation_loss": 0.3725208044052124,
      "epoch": 1.31,
      "learning_rate": 4.580485179064777e-05,
      "loss": 0.3704,
      "step": 1382,
      "task_loss": 0.350993275642395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.74158384234795,
      "compression_loss": 0.0,
      "distillation_loss": 0.03700024634599686,
      "epoch": 1.31,
      "learning_rate": 4.579894173289284e-05,
      "loss": 0.0392,
      "step": 1383,
      "task_loss": 0.0591222308576107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7416087249932148,
      "compression_loss": 0.0,
      "distillation_loss": 0.12477545440196991,
      "epoch": 1.31,
      "learning_rate": 4.579302789690997e-05,
      "loss": 0.1158,
      "step": 1384,
      "task_loss": 0.03471755236387253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7416336005715541,
      "compression_loss": 0.0,
      "distillation_loss": 0.1321507692337036,
      "epoch": 1.32,
      "learning_rate": 4.578711028377344e-05,
      "loss": 0.1349,
      "step": 1385,
      "task_loss": 0.1596865952014923
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7416584690839714,
      "compression_loss": 0.0,
      "distillation_loss": 0.09508243948221207,
      "epoch": 1.32,
      "learning_rate": 4.578118889455821e-05,
      "loss": 0.0901,
      "step": 1386,
      "task_loss": 0.04528198391199112
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7416833305314704,
      "compression_loss": 0.0,
      "distillation_loss": 0.09540075063705444,
      "epoch": 1.32,
      "learning_rate": 4.577526373033994e-05,
      "loss": 0.108,
      "step": 1387,
      "task_loss": 0.22137771546840668
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.741708184915055,
      "compression_loss": 0.0,
      "distillation_loss": 0.14678335189819336,
      "epoch": 1.32,
      "learning_rate": 4.576933479219496e-05,
      "loss": 0.1414,
      "step": 1388,
      "task_loss": 0.0929180309176445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7417330322357285,
      "compression_loss": 0.0,
      "distillation_loss": 0.18357060849666595,
      "epoch": 1.32,
      "learning_rate": 4.5763402081200294e-05,
      "loss": 0.1835,
      "step": 1389,
      "task_loss": 0.18299201130867004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7417578724944949,
      "compression_loss": 0.0,
      "distillation_loss": 0.18099266290664673,
      "epoch": 1.32,
      "learning_rate": 4.575746559843364e-05,
      "loss": 0.176,
      "step": 1390,
      "task_loss": 0.13100393116474152
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7417827056923577,
      "compression_loss": 0.0,
      "distillation_loss": 0.03351439908146858,
      "epoch": 1.32,
      "learning_rate": 4.5751525344973384e-05,
      "loss": 0.0415,
      "step": 1391,
      "task_loss": 0.11322241276502609
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7418075318303208,
      "compression_loss": 0.0,
      "distillation_loss": 0.06732058525085449,
      "epoch": 1.32,
      "learning_rate": 4.5745581321898615e-05,
      "loss": 0.0621,
      "step": 1392,
      "task_loss": 0.015335185453295708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7418323509093876,
      "compression_loss": 0.0,
      "distillation_loss": 0.2339862734079361,
      "epoch": 1.32,
      "learning_rate": 4.5739633530289085e-05,
      "loss": 0.2269,
      "step": 1393,
      "task_loss": 0.16297489404678345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.741857162930562,
      "compression_loss": 0.0,
      "distillation_loss": 0.042287249118089676,
      "epoch": 1.32,
      "learning_rate": 4.573368197122524e-05,
      "loss": 0.0504,
      "step": 1394,
      "task_loss": 0.12305901199579239
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7418819678948476,
      "compression_loss": 0.0,
      "distillation_loss": 0.20534148812294006,
      "epoch": 1.32,
      "learning_rate": 4.572772664578821e-05,
      "loss": 0.1949,
      "step": 1395,
      "task_loss": 0.10102957487106323
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.741906765803248,
      "compression_loss": 0.0,
      "distillation_loss": 0.056711532175540924,
      "epoch": 1.33,
      "learning_rate": 4.572176755505981e-05,
      "loss": 0.0517,
      "step": 1396,
      "task_loss": 0.006524372845888138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.741931556656767,
      "compression_loss": 0.0,
      "distillation_loss": 0.04466833174228668,
      "epoch": 1.33,
      "learning_rate": 4.571580470012254e-05,
      "loss": 0.0415,
      "step": 1397,
      "task_loss": 0.013324148952960968
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7419563404564083,
      "compression_loss": 0.0,
      "distillation_loss": 0.1416427046060562,
      "epoch": 1.33,
      "learning_rate": 4.5709838082059574e-05,
      "loss": 0.1331,
      "step": 1398,
      "task_loss": 0.05632413551211357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7419811172031756,
      "compression_loss": 0.0,
      "distillation_loss": 0.1042378842830658,
      "epoch": 1.33,
      "learning_rate": 4.570386770195478e-05,
      "loss": 0.0981,
      "step": 1399,
      "task_loss": 0.04313955828547478
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7420058868980723,
      "compression_loss": 0.0,
      "distillation_loss": 0.1780770719051361,
      "epoch": 1.33,
      "learning_rate": 4.569789356089271e-05,
      "loss": 0.177,
      "step": 1400,
      "task_loss": 0.16733302175998688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7420306495421025,
      "compression_loss": 0.0,
      "distillation_loss": 0.258390873670578,
      "epoch": 1.33,
      "learning_rate": 4.569191565995859e-05,
      "loss": 0.2603,
      "step": 1401,
      "task_loss": 0.2771160304546356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7420554051362696,
      "compression_loss": 0.0,
      "distillation_loss": 0.3139894902706146,
      "epoch": 1.33,
      "learning_rate": 4.568593400023834e-05,
      "loss": 0.3055,
      "step": 1402,
      "task_loss": 0.22894920408725739
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7420801536815774,
      "compression_loss": 0.0,
      "distillation_loss": 0.17985640466213226,
      "epoch": 1.33,
      "learning_rate": 4.567994858281855e-05,
      "loss": 0.1712,
      "step": 1403,
      "task_loss": 0.09336966276168823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7421048951790294,
      "compression_loss": 0.0,
      "distillation_loss": 0.19855880737304688,
      "epoch": 1.33,
      "learning_rate": 4.56739594087865e-05,
      "loss": 0.2139,
      "step": 1404,
      "task_loss": 0.35179561376571655
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7421296296296296,
      "compression_loss": 0.0,
      "distillation_loss": 0.4204038977622986,
      "epoch": 1.33,
      "learning_rate": 4.566796647923017e-05,
      "loss": 0.4184,
      "step": 1405,
      "task_loss": 0.4005380868911743
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7421543570343815,
      "compression_loss": 0.0,
      "distillation_loss": 0.18849390745162964,
      "epoch": 1.34,
      "learning_rate": 4.566196979523818e-05,
      "loss": 0.1992,
      "step": 1406,
      "task_loss": 0.2956838607788086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7421790773942887,
      "compression_loss": 0.0,
      "distillation_loss": 0.39693641662597656,
      "epoch": 1.34,
      "learning_rate": 4.5655969357899874e-05,
      "loss": 0.3956,
      "step": 1407,
      "task_loss": 0.38323667645454407
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.742203790710355,
      "compression_loss": 0.0,
      "distillation_loss": 0.21428659558296204,
      "epoch": 1.34,
      "learning_rate": 4.564996516830525e-05,
      "loss": 0.2072,
      "step": 1408,
      "task_loss": 0.14346663653850555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7422284969835841,
      "compression_loss": 0.0,
      "distillation_loss": 0.15829120576381683,
      "epoch": 1.34,
      "learning_rate": 4.564395722754501e-05,
      "loss": 0.1601,
      "step": 1409,
      "task_loss": 0.17646542191505432
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7422531962149795,
      "compression_loss": 0.0,
      "distillation_loss": 0.13824407756328583,
      "epoch": 1.34,
      "learning_rate": 4.56379455367105e-05,
      "loss": 0.129,
      "step": 1410,
      "task_loss": 0.04564966261386871
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7422778884055451,
      "compression_loss": 0.0,
      "distillation_loss": 0.05707935988903046,
      "epoch": 1.34,
      "learning_rate": 4.563193009689381e-05,
      "loss": 0.0538,
      "step": 1411,
      "task_loss": 0.023959307000041008
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7423025735562846,
      "compression_loss": 0.0,
      "distillation_loss": 0.42523133754730225,
      "epoch": 1.34,
      "learning_rate": 4.562591090918764e-05,
      "loss": 0.4085,
      "step": 1412,
      "task_loss": 0.2577897012233734
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7423272516682015,
      "compression_loss": 0.0,
      "distillation_loss": 0.37662801146507263,
      "epoch": 1.34,
      "learning_rate": 4.561988797468542e-05,
      "loss": 0.3691,
      "step": 1413,
      "task_loss": 0.30150848627090454
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7423519227422994,
      "compression_loss": 0.0,
      "distillation_loss": 0.21535269916057587,
      "epoch": 1.34,
      "learning_rate": 4.561386129448125e-05,
      "loss": 0.205,
      "step": 1414,
      "task_loss": 0.11162030696868896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7423765867795824,
      "compression_loss": 0.0,
      "distillation_loss": 0.14912311732769012,
      "epoch": 1.34,
      "learning_rate": 4.5607830869669885e-05,
      "loss": 0.1563,
      "step": 1415,
      "task_loss": 0.22072558104991913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7424012437810538,
      "compression_loss": 0.0,
      "distillation_loss": 0.1851930022239685,
      "epoch": 1.34,
      "learning_rate": 4.560179670134681e-05,
      "loss": 0.1752,
      "step": 1416,
      "task_loss": 0.08566315472126007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7424258937477174,
      "compression_loss": 0.0,
      "distillation_loss": 0.14120402932167053,
      "epoch": 1.35,
      "learning_rate": 4.559575879060813e-05,
      "loss": 0.1323,
      "step": 1417,
      "task_loss": 0.05225841701030731
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7424505366805769,
      "compression_loss": 0.0,
      "distillation_loss": 0.29736214876174927,
      "epoch": 1.35,
      "learning_rate": 4.5589717138550685e-05,
      "loss": 0.2882,
      "step": 1418,
      "task_loss": 0.20604415237903595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.742475172580636,
      "compression_loss": 0.0,
      "distillation_loss": 0.08903981745243073,
      "epoch": 1.35,
      "learning_rate": 4.5583671746271964e-05,
      "loss": 0.0856,
      "step": 1419,
      "task_loss": 0.054539553821086884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7424998014488984,
      "compression_loss": 0.0,
      "distillation_loss": 0.0935673713684082,
      "epoch": 1.35,
      "learning_rate": 4.557762261487013e-05,
      "loss": 0.0914,
      "step": 1420,
      "task_loss": 0.07140633463859558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7425244232863677,
      "compression_loss": 0.0,
      "distillation_loss": 0.19990570843219757,
      "epoch": 1.35,
      "learning_rate": 4.557156974544404e-05,
      "loss": 0.1898,
      "step": 1421,
      "task_loss": 0.09906322509050369
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7425490380940476,
      "compression_loss": 0.0,
      "distillation_loss": 0.24201074242591858,
      "epoch": 1.35,
      "learning_rate": 4.5565513139093244e-05,
      "loss": 0.2452,
      "step": 1422,
      "task_loss": 0.27348941564559937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7425736458729418,
      "compression_loss": 0.0,
      "distillation_loss": 0.2360510677099228,
      "epoch": 1.35,
      "learning_rate": 4.5559452796917936e-05,
      "loss": 0.2237,
      "step": 1423,
      "task_loss": 0.1127673014998436
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.742598246624054,
      "compression_loss": 0.0,
      "distillation_loss": 0.15818031132221222,
      "epoch": 1.35,
      "learning_rate": 4.555338872001901e-05,
      "loss": 0.1643,
      "step": 1424,
      "task_loss": 0.2189611941576004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7426228403483879,
      "compression_loss": 0.0,
      "distillation_loss": 0.45690128207206726,
      "epoch": 1.35,
      "learning_rate": 4.554732090949805e-05,
      "loss": 0.4419,
      "step": 1425,
      "task_loss": 0.30732840299606323
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7426474270469471,
      "compression_loss": 0.0,
      "distillation_loss": 0.15233120322227478,
      "epoch": 1.35,
      "learning_rate": 4.5541249366457276e-05,
      "loss": 0.1699,
      "step": 1426,
      "task_loss": 0.32829368114471436
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7426720067207354,
      "compression_loss": 0.0,
      "distillation_loss": 0.24415679275989532,
      "epoch": 1.36,
      "learning_rate": 4.5535174091999636e-05,
      "loss": 0.2323,
      "step": 1427,
      "task_loss": 0.12578970193862915
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7426965793707563,
      "compression_loss": 0.0,
      "distillation_loss": 0.10446126759052277,
      "epoch": 1.36,
      "learning_rate": 4.552909508722871e-05,
      "loss": 0.0981,
      "step": 1428,
      "task_loss": 0.040688008069992065
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7427211449980137,
      "compression_loss": 0.0,
      "distillation_loss": 0.13874554634094238,
      "epoch": 1.36,
      "learning_rate": 4.55230123532488e-05,
      "loss": 0.1403,
      "step": 1429,
      "task_loss": 0.15391620993614197
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7427457036035111,
      "compression_loss": 0.0,
      "distillation_loss": 0.04957931116223335,
      "epoch": 1.36,
      "learning_rate": 4.551692589116486e-05,
      "loss": 0.058,
      "step": 1430,
      "task_loss": 0.13414397835731506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7427702551882523,
      "compression_loss": 0.0,
      "distillation_loss": 0.18771857023239136,
      "epoch": 1.36,
      "learning_rate": 4.551083570208252e-05,
      "loss": 0.1799,
      "step": 1431,
      "task_loss": 0.10969773679971695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7427947997532409,
      "compression_loss": 0.0,
      "distillation_loss": 0.2879871129989624,
      "epoch": 1.36,
      "learning_rate": 4.550474178710809e-05,
      "loss": 0.2745,
      "step": 1432,
      "task_loss": 0.15289351344108582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7428193372994807,
      "compression_loss": 0.0,
      "distillation_loss": 0.0784875899553299,
      "epoch": 1.36,
      "learning_rate": 4.549864414734856e-05,
      "loss": 0.0731,
      "step": 1433,
      "task_loss": 0.02427070587873459
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7428438678279753,
      "compression_loss": 0.0,
      "distillation_loss": 0.08174806833267212,
      "epoch": 1.36,
      "learning_rate": 4.54925427839116e-05,
      "loss": 0.0832,
      "step": 1434,
      "task_loss": 0.09647224843502045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7428683913397283,
      "compression_loss": 0.0,
      "distillation_loss": 0.1861201375722885,
      "epoch": 1.36,
      "learning_rate": 4.548643769790556e-05,
      "loss": 0.1766,
      "step": 1435,
      "task_loss": 0.09077153354883194
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7428929078357436,
      "compression_loss": 0.0,
      "distillation_loss": 0.1611296832561493,
      "epoch": 1.36,
      "learning_rate": 4.548032889043944e-05,
      "loss": 0.169,
      "step": 1436,
      "task_loss": 0.2397400438785553
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7429174173170245,
      "compression_loss": 0.0,
      "distillation_loss": 0.14555801451206207,
      "epoch": 1.36,
      "learning_rate": 4.547421636262294e-05,
      "loss": 0.1467,
      "step": 1437,
      "task_loss": 0.1570826917886734
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7429419197845752,
      "compression_loss": 0.0,
      "distillation_loss": 0.23441661894321442,
      "epoch": 1.37,
      "learning_rate": 4.546810011556644e-05,
      "loss": 0.2309,
      "step": 1438,
      "task_loss": 0.19945275783538818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.742966415239399,
      "compression_loss": 0.0,
      "distillation_loss": 0.1089128702878952,
      "epoch": 1.37,
      "learning_rate": 4.546198015038097e-05,
      "loss": 0.1146,
      "step": 1439,
      "task_loss": 0.16534297168254852
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7429909036824998,
      "compression_loss": 0.0,
      "distillation_loss": 0.2072802484035492,
      "epoch": 1.37,
      "learning_rate": 4.545585646817826e-05,
      "loss": 0.1989,
      "step": 1440,
      "task_loss": 0.12354776263237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7430153851148811,
      "compression_loss": 0.0,
      "distillation_loss": 0.3501192629337311,
      "epoch": 1.37,
      "learning_rate": 4.544972907007071e-05,
      "loss": 0.335,
      "step": 1441,
      "task_loss": 0.19924892485141754
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7430398595375466,
      "compression_loss": 0.0,
      "distillation_loss": 0.45086967945098877,
      "epoch": 1.37,
      "learning_rate": 4.544359795717139e-05,
      "loss": 0.4323,
      "step": 1442,
      "task_loss": 0.265533447265625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7430643269515002,
      "compression_loss": 0.0,
      "distillation_loss": 0.12026548385620117,
      "epoch": 1.37,
      "learning_rate": 4.543746313059404e-05,
      "loss": 0.1146,
      "step": 1443,
      "task_loss": 0.063462033867836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7430887873577453,
      "compression_loss": 0.0,
      "distillation_loss": 0.17728132009506226,
      "epoch": 1.37,
      "learning_rate": 4.5431324591453094e-05,
      "loss": 0.1698,
      "step": 1444,
      "task_loss": 0.10240338742733002
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7431132407572857,
      "compression_loss": 0.0,
      "distillation_loss": 0.3929315209388733,
      "epoch": 1.37,
      "learning_rate": 4.5425182340863626e-05,
      "loss": 0.3889,
      "step": 1445,
      "task_loss": 0.3523111641407013
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7431376871511252,
      "compression_loss": 0.0,
      "distillation_loss": 0.14356978237628937,
      "epoch": 1.37,
      "learning_rate": 4.541903637994142e-05,
      "loss": 0.1433,
      "step": 1446,
      "task_loss": 0.14127568900585175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7431621265402673,
      "compression_loss": 0.0,
      "distillation_loss": 0.18327973783016205,
      "epoch": 1.37,
      "learning_rate": 4.541288670980291e-05,
      "loss": 0.171,
      "step": 1447,
      "task_loss": 0.060883235186338425
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7431865589257157,
      "compression_loss": 0.0,
      "distillation_loss": 0.12613250315189362,
      "epoch": 1.38,
      "learning_rate": 4.540673333156523e-05,
      "loss": 0.1266,
      "step": 1448,
      "task_loss": 0.1305573284626007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7432109843084742,
      "compression_loss": 0.0,
      "distillation_loss": 0.11743993312120438,
      "epoch": 1.38,
      "learning_rate": 4.540057624634616e-05,
      "loss": 0.11,
      "step": 1449,
      "task_loss": 0.04330931603908539
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7432354026895465,
      "compression_loss": 0.0,
      "distillation_loss": 0.16623345017433167,
      "epoch": 1.38,
      "learning_rate": 4.5394415455264164e-05,
      "loss": 0.1572,
      "step": 1450,
      "task_loss": 0.07557562738656998
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.743259814069936,
      "compression_loss": 0.0,
      "distillation_loss": 0.20014873147010803,
      "epoch": 1.38,
      "learning_rate": 4.538825095943838e-05,
      "loss": 0.2043,
      "step": 1451,
      "task_loss": 0.24151219427585602
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7432842184506467,
      "compression_loss": 0.0,
      "distillation_loss": 0.14636574685573578,
      "epoch": 1.38,
      "learning_rate": 4.538208275998861e-05,
      "loss": 0.1444,
      "step": 1452,
      "task_loss": 0.12678012251853943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7433086158326822,
      "compression_loss": 0.0,
      "distillation_loss": 0.12708376348018646,
      "epoch": 1.38,
      "learning_rate": 4.537591085803535e-05,
      "loss": 0.1272,
      "step": 1453,
      "task_loss": 0.1282857060432434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7433330062170461,
      "compression_loss": 0.0,
      "distillation_loss": 0.18036913871765137,
      "epoch": 1.38,
      "learning_rate": 4.5369735254699754e-05,
      "loss": 0.1817,
      "step": 1454,
      "task_loss": 0.19413068890571594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7433573896047422,
      "compression_loss": 0.0,
      "distillation_loss": 0.13496941328048706,
      "epoch": 1.38,
      "learning_rate": 4.536355595110365e-05,
      "loss": 0.1332,
      "step": 1455,
      "task_loss": 0.11723415553569794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.743381765996774,
      "compression_loss": 0.0,
      "distillation_loss": 0.19335393607616425,
      "epoch": 1.38,
      "learning_rate": 4.5357372948369534e-05,
      "loss": 0.1765,
      "step": 1456,
      "task_loss": 0.024988338351249695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7434061353941454,
      "compression_loss": 0.0,
      "distillation_loss": 0.19088754057884216,
      "epoch": 1.38,
      "learning_rate": 4.535118624762057e-05,
      "loss": 0.1783,
      "step": 1457,
      "task_loss": 0.06504514068365097
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7434304977978599,
      "compression_loss": 0.0,
      "distillation_loss": 0.059609562158584595,
      "epoch": 1.38,
      "learning_rate": 4.534499584998062e-05,
      "loss": 0.0554,
      "step": 1458,
      "task_loss": 0.01706998609006405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7434548532089212,
      "compression_loss": 0.0,
      "distillation_loss": 0.14300726354122162,
      "epoch": 1.39,
      "learning_rate": 4.533880175657419e-05,
      "loss": 0.1359,
      "step": 1459,
      "task_loss": 0.07233646512031555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7434792016283331,
      "compression_loss": 0.0,
      "distillation_loss": 0.07946481555700302,
      "epoch": 1.39,
      "learning_rate": 4.533260396852646e-05,
      "loss": 0.0883,
      "step": 1460,
      "task_loss": 0.167746901512146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7435035430570993,
      "compression_loss": 0.0,
      "distillation_loss": 0.16639113426208496,
      "epoch": 1.39,
      "learning_rate": 4.532640248696331e-05,
      "loss": 0.1619,
      "step": 1461,
      "task_loss": 0.12190359830856323
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7435278774962234,
      "compression_loss": 0.0,
      "distillation_loss": 0.1874895542860031,
      "epoch": 1.39,
      "learning_rate": 4.532019731301125e-05,
      "loss": 0.18,
      "step": 1462,
      "task_loss": 0.11228100210428238
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.743552204946709,
      "compression_loss": 0.0,
      "distillation_loss": 0.18930573761463165,
      "epoch": 1.39,
      "learning_rate": 4.531398844779749e-05,
      "loss": 0.189,
      "step": 1463,
      "task_loss": 0.18576987087726593
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7435765254095599,
      "compression_loss": 0.0,
      "distillation_loss": 0.17121566832065582,
      "epoch": 1.39,
      "learning_rate": 4.530777589244989e-05,
      "loss": 0.1633,
      "step": 1464,
      "task_loss": 0.09190039336681366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7436008388857798,
      "compression_loss": 0.0,
      "distillation_loss": 0.05851981043815613,
      "epoch": 1.39,
      "learning_rate": 4.5301559648096995e-05,
      "loss": 0.0584,
      "step": 1465,
      "task_loss": 0.05738149955868721
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7436251453763723,
      "compression_loss": 0.0,
      "distillation_loss": 0.18863308429718018,
      "epoch": 1.39,
      "learning_rate": 4.529533971586802e-05,
      "loss": 0.1878,
      "step": 1466,
      "task_loss": 0.1802268773317337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7436494448823411,
      "compression_loss": 0.0,
      "distillation_loss": 0.07734081149101257,
      "epoch": 1.39,
      "learning_rate": 4.5289116096892834e-05,
      "loss": 0.0851,
      "step": 1467,
      "task_loss": 0.15495401620864868
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7436737374046899,
      "compression_loss": 0.0,
      "distillation_loss": 0.07386526465415955,
      "epoch": 1.39,
      "learning_rate": 4.5282888792302e-05,
      "loss": 0.0687,
      "step": 1468,
      "task_loss": 0.022300483658909798
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7436980229444223,
      "compression_loss": 0.0,
      "distillation_loss": 0.19114729762077332,
      "epoch": 1.4,
      "learning_rate": 4.527665780322674e-05,
      "loss": 0.2027,
      "step": 1469,
      "task_loss": 0.3064166009426117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7437223015025421,
      "compression_loss": 0.0,
      "distillation_loss": 0.3061857223510742,
      "epoch": 1.4,
      "learning_rate": 4.527042313079893e-05,
      "loss": 0.2977,
      "step": 1470,
      "task_loss": 0.22168110311031342
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.743746573080053,
      "compression_loss": 0.0,
      "distillation_loss": 0.28121721744537354,
      "epoch": 1.4,
      "learning_rate": 4.526418477615114e-05,
      "loss": 0.2646,
      "step": 1471,
      "task_loss": 0.11492887139320374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7437708376779586,
      "compression_loss": 0.0,
      "distillation_loss": 0.12805329263210297,
      "epoch": 1.4,
      "learning_rate": 4.525794274041658e-05,
      "loss": 0.1237,
      "step": 1472,
      "task_loss": 0.08421643078327179
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7437950952972626,
      "compression_loss": 0.0,
      "distillation_loss": 0.21808746457099915,
      "epoch": 1.4,
      "learning_rate": 4.5251697024729165e-05,
      "loss": 0.2269,
      "step": 1473,
      "task_loss": 0.30643704533576965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7438193459389687,
      "compression_loss": 0.0,
      "distillation_loss": 0.12836065888404846,
      "epoch": 1.4,
      "learning_rate": 4.524544763022346e-05,
      "loss": 0.1309,
      "step": 1474,
      "task_loss": 0.15339140594005585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7438435896040806,
      "compression_loss": 0.0,
      "distillation_loss": 0.25501298904418945,
      "epoch": 1.4,
      "learning_rate": 4.523919455803468e-05,
      "loss": 0.2459,
      "step": 1475,
      "task_loss": 0.1641104519367218
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7438678262936019,
      "compression_loss": 0.0,
      "distillation_loss": 0.16674241423606873,
      "epoch": 1.4,
      "learning_rate": 4.5232937809298734e-05,
      "loss": 0.1652,
      "step": 1476,
      "task_loss": 0.1514662802219391
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7438920560085364,
      "compression_loss": 0.0,
      "distillation_loss": 0.18036165833473206,
      "epoch": 1.4,
      "learning_rate": 4.5226677385152206e-05,
      "loss": 0.1784,
      "step": 1477,
      "task_loss": 0.1612229347229004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7439162787498876,
      "compression_loss": 0.0,
      "distillation_loss": 0.05593033879995346,
      "epoch": 1.4,
      "learning_rate": 4.522041328673231e-05,
      "loss": 0.0531,
      "step": 1478,
      "task_loss": 0.027741387486457825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7439404945186594,
      "compression_loss": 0.0,
      "distillation_loss": 0.07034408301115036,
      "epoch": 1.4,
      "learning_rate": 4.521414551517695e-05,
      "loss": 0.0644,
      "step": 1479,
      "task_loss": 0.011257486417889595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7439647033158554,
      "compression_loss": 0.0,
      "distillation_loss": 0.2267724722623825,
      "epoch": 1.41,
      "learning_rate": 4.520787407162471e-05,
      "loss": 0.2152,
      "step": 1480,
      "task_loss": 0.11058557778596878
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7439889051424792,
      "compression_loss": 0.0,
      "distillation_loss": 0.07127425074577332,
      "epoch": 1.41,
      "learning_rate": 4.520159895721483e-05,
      "loss": 0.0828,
      "step": 1481,
      "task_loss": 0.18699173629283905
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7440130999995346,
      "compression_loss": 0.0,
      "distillation_loss": 0.1930515021085739,
      "epoch": 1.41,
      "learning_rate": 4.51953201730872e-05,
      "loss": 0.1903,
      "step": 1482,
      "task_loss": 0.1656961441040039
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7440372878880251,
      "compression_loss": 0.0,
      "distillation_loss": 0.15473680198192596,
      "epoch": 1.41,
      "learning_rate": 4.51890377203824e-05,
      "loss": 0.145,
      "step": 1483,
      "task_loss": 0.05734732747077942
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7440614688089547,
      "compression_loss": 0.0,
      "distillation_loss": 0.22360102832317352,
      "epoch": 1.41,
      "learning_rate": 4.518275160024167e-05,
      "loss": 0.2159,
      "step": 1484,
      "task_loss": 0.14627224206924438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7440856427633268,
      "compression_loss": 0.0,
      "distillation_loss": 0.17431600391864777,
      "epoch": 1.41,
      "learning_rate": 4.5176461813806904e-05,
      "loss": 0.1683,
      "step": 1485,
      "task_loss": 0.1145024299621582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7441098097521451,
      "compression_loss": 0.0,
      "distillation_loss": 0.14846046268939972,
      "epoch": 1.41,
      "learning_rate": 4.5170168362220686e-05,
      "loss": 0.1464,
      "step": 1486,
      "task_loss": 0.12777863442897797
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7441339697764136,
      "compression_loss": 0.0,
      "distillation_loss": 0.11251084506511688,
      "epoch": 1.41,
      "learning_rate": 4.516387124662624e-05,
      "loss": 0.1045,
      "step": 1487,
      "task_loss": 0.032140232622623444
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7441581228371354,
      "compression_loss": 0.0,
      "distillation_loss": 0.29036539793014526,
      "epoch": 1.41,
      "learning_rate": 4.5157570468167464e-05,
      "loss": 0.2763,
      "step": 1488,
      "task_loss": 0.14925380051136017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7441822689353148,
      "compression_loss": 0.0,
      "distillation_loss": 0.06524545699357986,
      "epoch": 1.41,
      "learning_rate": 4.5151266027988946e-05,
      "loss": 0.0624,
      "step": 1489,
      "task_loss": 0.03709521144628525
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.744206408071955,
      "compression_loss": 0.0,
      "distillation_loss": 0.12188740074634552,
      "epoch": 1.42,
      "learning_rate": 4.51449579272359e-05,
      "loss": 0.1168,
      "step": 1490,
      "task_loss": 0.07146313041448593
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.74423054024806,
      "compression_loss": 0.0,
      "distillation_loss": 0.035563819110393524,
      "epoch": 1.42,
      "learning_rate": 4.5138646167054224e-05,
      "loss": 0.0455,
      "step": 1491,
      "task_loss": 0.1353863775730133
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7442546654646334,
      "compression_loss": 0.0,
      "distillation_loss": 0.18486768007278442,
      "epoch": 1.42,
      "learning_rate": 4.513233074859049e-05,
      "loss": 0.1757,
      "step": 1492,
      "task_loss": 0.09314113855361938
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7442787837226787,
      "compression_loss": 0.0,
      "distillation_loss": 0.11438624560832977,
      "epoch": 1.42,
      "learning_rate": 4.512601167299191e-05,
      "loss": 0.1168,
      "step": 1493,
      "task_loss": 0.13881582021713257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7443028950231999,
      "compression_loss": 0.0,
      "distillation_loss": 0.03467051684856415,
      "epoch": 1.42,
      "learning_rate": 4.511968894140639e-05,
      "loss": 0.0348,
      "step": 1494,
      "task_loss": 0.0362866148352623
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7443269993672005,
      "compression_loss": 0.0,
      "distillation_loss": 0.15380631387233734,
      "epoch": 1.42,
      "learning_rate": 4.511336255498247e-05,
      "loss": 0.1452,
      "step": 1495,
      "task_loss": 0.06820768862962723
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7443510967556841,
      "compression_loss": 0.0,
      "distillation_loss": 0.057968005537986755,
      "epoch": 1.42,
      "learning_rate": 4.510703251486937e-05,
      "loss": 0.0636,
      "step": 1496,
      "task_loss": 0.11408950388431549
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7443751871896545,
      "compression_loss": 0.0,
      "distillation_loss": 0.1298018991947174,
      "epoch": 1.42,
      "learning_rate": 4.5100698822216984e-05,
      "loss": 0.1205,
      "step": 1497,
      "task_loss": 0.03705129399895668
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7443992706701154,
      "compression_loss": 0.0,
      "distillation_loss": 0.056667059659957886,
      "epoch": 1.42,
      "learning_rate": 4.509436147817585e-05,
      "loss": 0.0534,
      "step": 1498,
      "task_loss": 0.02439264766871929
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7444233471980704,
      "compression_loss": 0.0,
      "distillation_loss": 0.123785100877285,
      "epoch": 1.42,
      "learning_rate": 4.5088020483897184e-05,
      "loss": 0.1167,
      "step": 1499,
      "task_loss": 0.05303073301911354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7444474167745232,
      "compression_loss": 0.0,
      "distillation_loss": 0.02330487221479416,
      "epoch": 1.42,
      "learning_rate": 4.508167584053285e-05,
      "loss": 0.0425,
      "step": 1500,
      "task_loss": 0.21552759408950806
    },
    {
      "epoch": 1.42,
      "eval_accuracy": 0.911697247706422,
      "eval_loss": 0.38651296496391296,
      "eval_runtime": 18.7646,
      "eval_samples_per_second": 46.471,
      "eval_steps_per_second": 5.809,
      "step": 1500
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7444714794004775,
      "compression_loss": 0.0,
      "distillation_loss": 0.11092271655797958,
      "epoch": 1.43,
      "learning_rate": 4.507532754923537e-05,
      "loss": 0.1073,
      "step": 1501,
      "task_loss": 0.07452307641506195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7444955350769371,
      "compression_loss": 0.0,
      "distillation_loss": 0.06650504469871521,
      "epoch": 1.43,
      "learning_rate": 4.506897561115797e-05,
      "loss": 0.0621,
      "step": 1502,
      "task_loss": 0.02240423485636711
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7445195838049055,
      "compression_loss": 0.0,
      "distillation_loss": 0.16672199964523315,
      "epoch": 1.43,
      "learning_rate": 4.506262002745449e-05,
      "loss": 0.1695,
      "step": 1503,
      "task_loss": 0.194443479180336
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7445436255853864,
      "compression_loss": 0.0,
      "distillation_loss": 0.07140068709850311,
      "epoch": 1.43,
      "learning_rate": 4.505626079927947e-05,
      "loss": 0.0655,
      "step": 1504,
      "task_loss": 0.012316873297095299
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7445676604193836,
      "compression_loss": 0.0,
      "distillation_loss": 0.21132929623126984,
      "epoch": 1.43,
      "learning_rate": 4.504989792778808e-05,
      "loss": 0.2003,
      "step": 1505,
      "task_loss": 0.10102443397045135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7445916883079007,
      "compression_loss": 0.0,
      "distillation_loss": 0.1785595715045929,
      "epoch": 1.43,
      "learning_rate": 4.504353141413616e-05,
      "loss": 0.1787,
      "step": 1506,
      "task_loss": 0.18019554018974304
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7446157092519414,
      "compression_loss": 0.0,
      "distillation_loss": 0.25665420293807983,
      "epoch": 1.43,
      "learning_rate": 4.5037161259480246e-05,
      "loss": 0.2455,
      "step": 1507,
      "task_loss": 0.14560416340827942
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7446397232525094,
      "compression_loss": 0.0,
      "distillation_loss": 0.04967069625854492,
      "epoch": 1.43,
      "learning_rate": 4.5030787464977476e-05,
      "loss": 0.0577,
      "step": 1508,
      "task_loss": 0.13003495335578918
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7446637303106083,
      "compression_loss": 0.0,
      "distillation_loss": 0.17093107104301453,
      "epoch": 1.43,
      "learning_rate": 4.50244100317857e-05,
      "loss": 0.171,
      "step": 1509,
      "task_loss": 0.17199328541755676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7446877304272419,
      "compression_loss": 0.0,
      "distillation_loss": 0.06911082565784454,
      "epoch": 1.43,
      "learning_rate": 4.5018028961063394e-05,
      "loss": 0.0759,
      "step": 1510,
      "task_loss": 0.13733558356761932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7447117236034139,
      "compression_loss": 0.0,
      "distillation_loss": 0.21413826942443848,
      "epoch": 1.43,
      "learning_rate": 4.501164425396973e-05,
      "loss": 0.2076,
      "step": 1511,
      "task_loss": 0.1490751951932907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7447357098401277,
      "compression_loss": 0.0,
      "distillation_loss": 0.1935446858406067,
      "epoch": 1.44,
      "learning_rate": 4.5005255911664507e-05,
      "loss": 0.1988,
      "step": 1512,
      "task_loss": 0.2461317479610443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7447596891383874,
      "compression_loss": 0.0,
      "distillation_loss": 0.1821810007095337,
      "epoch": 1.44,
      "learning_rate": 4.49988639353082e-05,
      "loss": 0.1794,
      "step": 1513,
      "task_loss": 0.15411117672920227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7447836614991964,
      "compression_loss": 0.0,
      "distillation_loss": 0.16811317205429077,
      "epoch": 1.44,
      "learning_rate": 4.4992468326061944e-05,
      "loss": 0.1679,
      "step": 1514,
      "task_loss": 0.16646404564380646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7448076269235584,
      "compression_loss": 0.0,
      "distillation_loss": 0.13295818865299225,
      "epoch": 1.44,
      "learning_rate": 4.498606908508754e-05,
      "loss": 0.1242,
      "step": 1515,
      "task_loss": 0.045377686619758606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7448315854124772,
      "compression_loss": 0.0,
      "distillation_loss": 0.16021104156970978,
      "epoch": 1.44,
      "learning_rate": 4.4979666213547414e-05,
      "loss": 0.1632,
      "step": 1516,
      "task_loss": 0.19016344845294952
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7448555369669564,
      "compression_loss": 0.0,
      "distillation_loss": 0.1837073564529419,
      "epoch": 1.44,
      "learning_rate": 4.497325971260471e-05,
      "loss": 0.1772,
      "step": 1517,
      "task_loss": 0.11903562396764755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7448794815879997,
      "compression_loss": 0.0,
      "distillation_loss": 0.08035746216773987,
      "epoch": 1.44,
      "learning_rate": 4.496684958342319e-05,
      "loss": 0.076,
      "step": 1518,
      "task_loss": 0.03652361035346985
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7449034192766107,
      "compression_loss": 0.0,
      "distillation_loss": 0.04287096485495567,
      "epoch": 1.44,
      "learning_rate": 4.4960435827167266e-05,
      "loss": 0.0395,
      "step": 1519,
      "task_loss": 0.009310789406299591
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7449273500337933,
      "compression_loss": 0.0,
      "distillation_loss": 0.04646936431527138,
      "epoch": 1.44,
      "learning_rate": 4.495401844500205e-05,
      "loss": 0.0439,
      "step": 1520,
      "task_loss": 0.02119472809135914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.744951273860551,
      "compression_loss": 0.0,
      "distillation_loss": 0.20926059782505035,
      "epoch": 1.44,
      "learning_rate": 4.494759743809329e-05,
      "loss": 0.2045,
      "step": 1521,
      "task_loss": 0.16188855469226837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7449751907578875,
      "compression_loss": 0.0,
      "distillation_loss": 0.11345314979553223,
      "epoch": 1.45,
      "learning_rate": 4.494117280760739e-05,
      "loss": 0.1181,
      "step": 1522,
      "task_loss": 0.16004882752895355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7449991007268065,
      "compression_loss": 0.0,
      "distillation_loss": 0.16868895292282104,
      "epoch": 1.45,
      "learning_rate": 4.49347445547114e-05,
      "loss": 0.1641,
      "step": 1523,
      "task_loss": 0.12286947667598724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7450230037683118,
      "compression_loss": 0.0,
      "distillation_loss": 0.17652767896652222,
      "epoch": 1.45,
      "learning_rate": 4.4928312680573064e-05,
      "loss": 0.168,
      "step": 1524,
      "task_loss": 0.09076812118291855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7450468998834069,
      "compression_loss": 0.0,
      "distillation_loss": 0.05617145821452141,
      "epoch": 1.45,
      "learning_rate": 4.492187718636075e-05,
      "loss": 0.0604,
      "step": 1525,
      "task_loss": 0.09820869565010071
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7450707890730955,
      "compression_loss": 0.0,
      "distillation_loss": 0.216825470328331,
      "epoch": 1.45,
      "learning_rate": 4.49154380732435e-05,
      "loss": 0.2122,
      "step": 1526,
      "task_loss": 0.1709357053041458
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7450946713383815,
      "compression_loss": 0.0,
      "distillation_loss": 0.4693998694419861,
      "epoch": 1.45,
      "learning_rate": 4.490899534239101e-05,
      "loss": 0.4483,
      "step": 1527,
      "task_loss": 0.2588621973991394
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7451185466802683,
      "compression_loss": 0.0,
      "distillation_loss": 0.2971939742565155,
      "epoch": 1.45,
      "learning_rate": 4.490254899497364e-05,
      "loss": 0.2901,
      "step": 1528,
      "task_loss": 0.22666522860527039
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7451424150997598,
      "compression_loss": 0.0,
      "distillation_loss": 0.25744807720184326,
      "epoch": 1.45,
      "learning_rate": 4.4896099032162386e-05,
      "loss": 0.2487,
      "step": 1529,
      "task_loss": 0.17000789940357208
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7451662765978594,
      "compression_loss": 0.0,
      "distillation_loss": 0.13210158050060272,
      "epoch": 1.45,
      "learning_rate": 4.488964545512892e-05,
      "loss": 0.1324,
      "step": 1530,
      "task_loss": 0.13510762155056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7451901311755711,
      "compression_loss": 0.0,
      "distillation_loss": 0.2513887882232666,
      "epoch": 1.45,
      "learning_rate": 4.488318826504557e-05,
      "loss": 0.2502,
      "step": 1531,
      "task_loss": 0.23999468982219696
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7452139788338985,
      "compression_loss": 0.0,
      "distillation_loss": 0.2592649757862091,
      "epoch": 1.45,
      "learning_rate": 4.4876727463085324e-05,
      "loss": 0.2465,
      "step": 1532,
      "task_loss": 0.13125677406787872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7452378195738452,
      "compression_loss": 0.0,
      "distillation_loss": 0.26413312554359436,
      "epoch": 1.46,
      "learning_rate": 4.487026305042179e-05,
      "loss": 0.2514,
      "step": 1533,
      "task_loss": 0.13711029291152954
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7452616533964148,
      "compression_loss": 0.0,
      "distillation_loss": 0.3824172019958496,
      "epoch": 1.46,
      "learning_rate": 4.4863795028229286e-05,
      "loss": 0.3723,
      "step": 1534,
      "task_loss": 0.281066358089447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7452854803026112,
      "compression_loss": 0.0,
      "distillation_loss": 0.22528879344463348,
      "epoch": 1.46,
      "learning_rate": 4.4857323397682746e-05,
      "loss": 0.2242,
      "step": 1535,
      "task_loss": 0.21392083168029785
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.745309300293438,
      "compression_loss": 0.0,
      "distillation_loss": 0.12722215056419373,
      "epoch": 1.46,
      "learning_rate": 4.485084815995778e-05,
      "loss": 0.1359,
      "step": 1536,
      "task_loss": 0.2140396535396576
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7453331133698988,
      "compression_loss": 0.0,
      "distillation_loss": 0.14468204975128174,
      "epoch": 1.46,
      "learning_rate": 4.484436931623064e-05,
      "loss": 0.1436,
      "step": 1537,
      "task_loss": 0.1337466537952423
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7453569195329973,
      "compression_loss": 0.0,
      "distillation_loss": 0.41913074254989624,
      "epoch": 1.46,
      "learning_rate": 4.4837886867678245e-05,
      "loss": 0.4061,
      "step": 1538,
      "task_loss": 0.28871697187423706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7453807187837374,
      "compression_loss": 0.0,
      "distillation_loss": 0.12353764474391937,
      "epoch": 1.46,
      "learning_rate": 4.4831400815478164e-05,
      "loss": 0.117,
      "step": 1539,
      "task_loss": 0.05776692181825638
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7454045111231224,
      "compression_loss": 0.0,
      "distillation_loss": 0.07819847762584686,
      "epoch": 1.46,
      "learning_rate": 4.482491116080861e-05,
      "loss": 0.0901,
      "step": 1540,
      "task_loss": 0.19672957062721252
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7454282965521564,
      "compression_loss": 0.0,
      "distillation_loss": 0.3406957983970642,
      "epoch": 1.46,
      "learning_rate": 4.4818417904848466e-05,
      "loss": 0.3262,
      "step": 1541,
      "task_loss": 0.19611957669258118
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7454520750718426,
      "compression_loss": 0.0,
      "distillation_loss": 0.1422961801290512,
      "epoch": 1.46,
      "learning_rate": 4.481192104877726e-05,
      "loss": 0.1342,
      "step": 1542,
      "task_loss": 0.060906361788511276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7454758466831852,
      "compression_loss": 0.0,
      "distillation_loss": 0.05376805365085602,
      "epoch": 1.47,
      "learning_rate": 4.480542059377519e-05,
      "loss": 0.0542,
      "step": 1543,
      "task_loss": 0.05844269320368767
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7454996113871876,
      "compression_loss": 0.0,
      "distillation_loss": 0.20722997188568115,
      "epoch": 1.47,
      "learning_rate": 4.479891654102307e-05,
      "loss": 0.2045,
      "step": 1544,
      "task_loss": 0.1803072690963745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7455233691848534,
      "compression_loss": 0.0,
      "distillation_loss": 0.2142692506313324,
      "epoch": 1.47,
      "learning_rate": 4.4792408891702426e-05,
      "loss": 0.2035,
      "step": 1545,
      "task_loss": 0.10616156458854675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7455471200771865,
      "compression_loss": 0.0,
      "distillation_loss": 0.14973235130310059,
      "epoch": 1.47,
      "learning_rate": 4.4785897646995376e-05,
      "loss": 0.1475,
      "step": 1546,
      "task_loss": 0.12703794240951538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7455708640651905,
      "compression_loss": 0.0,
      "distillation_loss": 0.18545937538146973,
      "epoch": 1.47,
      "learning_rate": 4.477938280808473e-05,
      "loss": 0.184,
      "step": 1547,
      "task_loss": 0.17124545574188232
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.745594601149869,
      "compression_loss": 0.0,
      "distillation_loss": 0.1463180035352707,
      "epoch": 1.47,
      "learning_rate": 4.4772864376153936e-05,
      "loss": 0.1379,
      "step": 1548,
      "task_loss": 0.06201765686273575
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7456183313322257,
      "compression_loss": 0.0,
      "distillation_loss": 0.1488766223192215,
      "epoch": 1.47,
      "learning_rate": 4.4766342352387106e-05,
      "loss": 0.1424,
      "step": 1549,
      "task_loss": 0.08362394571304321
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7456420546132644,
      "compression_loss": 0.0,
      "distillation_loss": 0.09970419108867645,
      "epoch": 1.47,
      "learning_rate": 4.475981673796899e-05,
      "loss": 0.1015,
      "step": 1550,
      "task_loss": 0.11792711168527603
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7456657709939887,
      "compression_loss": 0.0,
      "distillation_loss": 0.17068204283714294,
      "epoch": 1.47,
      "learning_rate": 4.475328753408499e-05,
      "loss": 0.1719,
      "step": 1551,
      "task_loss": 0.18303748965263367
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7456894804754023,
      "compression_loss": 0.0,
      "distillation_loss": 0.05580803379416466,
      "epoch": 1.47,
      "learning_rate": 4.474675474192119e-05,
      "loss": 0.0535,
      "step": 1552,
      "task_loss": 0.03291773051023483
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7457131830585089,
      "compression_loss": 0.0,
      "distillation_loss": 0.033489570021629333,
      "epoch": 1.47,
      "learning_rate": 4.4740218362664276e-05,
      "loss": 0.0308,
      "step": 1553,
      "task_loss": 0.006132926791906357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.745736878744312,
      "compression_loss": 0.0,
      "distillation_loss": 0.0999932736158371,
      "epoch": 1.48,
      "learning_rate": 4.473367839750165e-05,
      "loss": 0.1088,
      "step": 1554,
      "task_loss": 0.18756511807441711
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7457605675338156,
      "compression_loss": 0.0,
      "distillation_loss": 0.1821468025445938,
      "epoch": 1.48,
      "learning_rate": 4.4727134847621276e-05,
      "loss": 0.174,
      "step": 1555,
      "task_loss": 0.1006765216588974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7457842494280232,
      "compression_loss": 0.0,
      "distillation_loss": 0.15182924270629883,
      "epoch": 1.48,
      "learning_rate": 4.4720587714211863e-05,
      "loss": 0.1564,
      "step": 1556,
      "task_loss": 0.1978447288274765
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7458079244279385,
      "compression_loss": 0.0,
      "distillation_loss": 0.28013527393341064,
      "epoch": 1.48,
      "learning_rate": 4.471403699846272e-05,
      "loss": 0.2684,
      "step": 1557,
      "task_loss": 0.16278645396232605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7458315925345651,
      "compression_loss": 0.0,
      "distillation_loss": 0.023632332682609558,
      "epoch": 1.48,
      "learning_rate": 4.470748270156381e-05,
      "loss": 0.0296,
      "step": 1558,
      "task_loss": 0.08337298780679703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7458552537489069,
      "compression_loss": 0.0,
      "distillation_loss": 0.07167325913906097,
      "epoch": 1.48,
      "learning_rate": 4.4700924824705745e-05,
      "loss": 0.0692,
      "step": 1559,
      "task_loss": 0.047288812696933746
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7458789080719673,
      "compression_loss": 0.0,
      "distillation_loss": 0.053809523582458496,
      "epoch": 1.48,
      "learning_rate": 4.469436336907982e-05,
      "loss": 0.0576,
      "step": 1560,
      "task_loss": 0.09203386306762695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7459025555047503,
      "compression_loss": 0.0,
      "distillation_loss": 0.08844821155071259,
      "epoch": 1.48,
      "learning_rate": 4.4687798335877936e-05,
      "loss": 0.088,
      "step": 1561,
      "task_loss": 0.08441342413425446
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7459261960482593,
      "compression_loss": 0.0,
      "distillation_loss": 0.053808607161045074,
      "epoch": 1.48,
      "learning_rate": 4.4681229726292664e-05,
      "loss": 0.0586,
      "step": 1562,
      "task_loss": 0.1021692305803299
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7459498297034981,
      "compression_loss": 0.0,
      "distillation_loss": 0.24225428700447083,
      "epoch": 1.48,
      "learning_rate": 4.4674657541517227e-05,
      "loss": 0.2405,
      "step": 1563,
      "task_loss": 0.22495496273040771
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7459734564714704,
      "compression_loss": 0.0,
      "distillation_loss": 0.1985190361738205,
      "epoch": 1.49,
      "learning_rate": 4.466808178274549e-05,
      "loss": 0.1868,
      "step": 1564,
      "task_loss": 0.08100724220275879
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7459970763531799,
      "compression_loss": 0.0,
      "distillation_loss": 0.2816087603569031,
      "epoch": 1.49,
      "learning_rate": 4.4661502451171975e-05,
      "loss": 0.2677,
      "step": 1565,
      "task_loss": 0.14259889721870422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7460206893496302,
      "compression_loss": 0.0,
      "distillation_loss": 0.16295886039733887,
      "epoch": 1.49,
      "learning_rate": 4.465491954799186e-05,
      "loss": 0.1618,
      "step": 1566,
      "task_loss": 0.1518009454011917
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.746044295461825,
      "compression_loss": 0.0,
      "distillation_loss": 0.17871344089508057,
      "epoch": 1.49,
      "learning_rate": 4.4648333074400936e-05,
      "loss": 0.171,
      "step": 1567,
      "task_loss": 0.10131989419460297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7460678946907681,
      "compression_loss": 0.0,
      "distillation_loss": 0.10870079696178436,
      "epoch": 1.49,
      "learning_rate": 4.464174303159569e-05,
      "loss": 0.1071,
      "step": 1568,
      "task_loss": 0.09272871166467667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.746091487037463,
      "compression_loss": 0.0,
      "distillation_loss": 0.11727038025856018,
      "epoch": 1.49,
      "learning_rate": 4.463514942077323e-05,
      "loss": 0.1156,
      "step": 1569,
      "task_loss": 0.10054733604192734
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7461150725029135,
      "compression_loss": 0.0,
      "distillation_loss": 0.03882730007171631,
      "epoch": 1.49,
      "learning_rate": 4.4628552243131304e-05,
      "loss": 0.0442,
      "step": 1570,
      "task_loss": 0.09213192760944366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7461386510881233,
      "compression_loss": 0.0,
      "distillation_loss": 0.09602994471788406,
      "epoch": 1.49,
      "learning_rate": 4.462195149986833e-05,
      "loss": 0.09,
      "step": 1571,
      "task_loss": 0.03555634990334511
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7461622227940959,
      "compression_loss": 0.0,
      "distillation_loss": 0.2854123115539551,
      "epoch": 1.49,
      "learning_rate": 4.4615347192183375e-05,
      "loss": 0.2702,
      "step": 1572,
      "task_loss": 0.13358475267887115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7461857876218352,
      "compression_loss": 0.0,
      "distillation_loss": 0.08520927280187607,
      "epoch": 1.49,
      "learning_rate": 4.4608739321276126e-05,
      "loss": 0.0797,
      "step": 1573,
      "task_loss": 0.03009282425045967
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7462093455723449,
      "compression_loss": 0.0,
      "distillation_loss": 0.074330173432827,
      "epoch": 1.49,
      "learning_rate": 4.4602127888346944e-05,
      "loss": 0.0698,
      "step": 1574,
      "task_loss": 0.028723739087581635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7462328966466284,
      "compression_loss": 0.0,
      "distillation_loss": 0.06121968850493431,
      "epoch": 1.5,
      "learning_rate": 4.459551289459684e-05,
      "loss": 0.0707,
      "step": 1575,
      "task_loss": 0.15602229535579681
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7462564408456897,
      "compression_loss": 0.0,
      "distillation_loss": 0.14412710070610046,
      "epoch": 1.5,
      "learning_rate": 4.4588894341227426e-05,
      "loss": 0.1395,
      "step": 1576,
      "task_loss": 0.09811600297689438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7462799781705323,
      "compression_loss": 0.0,
      "distillation_loss": 0.2897692918777466,
      "epoch": 1.5,
      "learning_rate": 4.4582272229441024e-05,
      "loss": 0.2797,
      "step": 1577,
      "task_loss": 0.1895175725221634
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7463035086221598,
      "compression_loss": 0.0,
      "distillation_loss": 0.06360229849815369,
      "epoch": 1.5,
      "learning_rate": 4.457564656044056e-05,
      "loss": 0.0622,
      "step": 1578,
      "task_loss": 0.04995894804596901
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7463270322015761,
      "compression_loss": 0.0,
      "distillation_loss": 0.0988512635231018,
      "epoch": 1.5,
      "learning_rate": 4.456901733542962e-05,
      "loss": 0.1002,
      "step": 1579,
      "task_loss": 0.11196555197238922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7463505489097848,
      "compression_loss": 0.0,
      "distillation_loss": 0.11553806811571121,
      "epoch": 1.5,
      "learning_rate": 4.4562384555612436e-05,
      "loss": 0.109,
      "step": 1580,
      "task_loss": 0.050435107201337814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7463740587477896,
      "compression_loss": 0.0,
      "distillation_loss": 0.4326043128967285,
      "epoch": 1.5,
      "learning_rate": 4.455574822219388e-05,
      "loss": 0.4232,
      "step": 1581,
      "task_loss": 0.3386293053627014
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7463975617165942,
      "compression_loss": 0.0,
      "distillation_loss": 0.2798699736595154,
      "epoch": 1.5,
      "learning_rate": 4.454910833637949e-05,
      "loss": 0.2735,
      "step": 1582,
      "task_loss": 0.21614964306354523
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7464210578172021,
      "compression_loss": 0.0,
      "distillation_loss": 0.1274246722459793,
      "epoch": 1.5,
      "learning_rate": 4.454246489937541e-05,
      "loss": 0.1217,
      "step": 1583,
      "task_loss": 0.07030543684959412
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7464445470506172,
      "compression_loss": 0.0,
      "distillation_loss": 0.07745787501335144,
      "epoch": 1.5,
      "learning_rate": 4.4535817912388466e-05,
      "loss": 0.081,
      "step": 1584,
      "task_loss": 0.11285798251628876
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7464680294178432,
      "compression_loss": 0.0,
      "distillation_loss": 0.2271667718887329,
      "epoch": 1.51,
      "learning_rate": 4.4529167376626116e-05,
      "loss": 0.2175,
      "step": 1585,
      "task_loss": 0.13050048053264618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7464915049198835,
      "compression_loss": 0.0,
      "distillation_loss": 0.3326779007911682,
      "epoch": 1.51,
      "learning_rate": 4.4522513293296456e-05,
      "loss": 0.3292,
      "step": 1586,
      "task_loss": 0.2977886497974396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7465149735577421,
      "compression_loss": 0.0,
      "distillation_loss": 0.04014962911605835,
      "epoch": 1.51,
      "learning_rate": 4.451585566360823e-05,
      "loss": 0.052,
      "step": 1587,
      "task_loss": 0.15873615443706512
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7465384353324225,
      "compression_loss": 0.0,
      "distillation_loss": 0.154592826962471,
      "epoch": 1.51,
      "learning_rate": 4.450919448877084e-05,
      "loss": 0.1459,
      "step": 1588,
      "task_loss": 0.06723834574222565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7465618902449284,
      "compression_loss": 0.0,
      "distillation_loss": 0.05010029301047325,
      "epoch": 1.51,
      "learning_rate": 4.4502529769994314e-05,
      "loss": 0.0464,
      "step": 1589,
      "task_loss": 0.012682372704148293
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7465853382962635,
      "compression_loss": 0.0,
      "distillation_loss": 0.08115085959434509,
      "epoch": 1.51,
      "learning_rate": 4.449586150848934e-05,
      "loss": 0.0893,
      "step": 1590,
      "task_loss": 0.1630755513906479
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7466087794874315,
      "compression_loss": 0.0,
      "distillation_loss": 0.21416254341602325,
      "epoch": 1.51,
      "learning_rate": 4.44891897054672e-05,
      "loss": 0.2044,
      "step": 1591,
      "task_loss": 0.11678683757781982
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7466322138194361,
      "compression_loss": 0.0,
      "distillation_loss": 0.17701666057109833,
      "epoch": 1.51,
      "learning_rate": 4.4482514362139915e-05,
      "loss": 0.1665,
      "step": 1592,
      "task_loss": 0.07226403057575226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.746655641293281,
      "compression_loss": 0.0,
      "distillation_loss": 0.09402104467153549,
      "epoch": 1.51,
      "learning_rate": 4.4475835479720065e-05,
      "loss": 0.0962,
      "step": 1593,
      "task_loss": 0.11615651845932007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7466790619099698,
      "compression_loss": 0.0,
      "distillation_loss": 0.3029029071331024,
      "epoch": 1.51,
      "learning_rate": 4.4469153059420895e-05,
      "loss": 0.2929,
      "step": 1594,
      "task_loss": 0.20262876152992249
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7467024756705063,
      "compression_loss": 0.0,
      "distillation_loss": 0.17734010517597198,
      "epoch": 1.51,
      "learning_rate": 4.4462467102456305e-05,
      "loss": 0.1858,
      "step": 1595,
      "task_loss": 0.26218533515930176
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.746725882575894,
      "compression_loss": 0.0,
      "distillation_loss": 0.24488165974617004,
      "epoch": 1.52,
      "learning_rate": 4.4455777610040846e-05,
      "loss": 0.252,
      "step": 1596,
      "task_loss": 0.3160625696182251
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7467492826271367,
      "compression_loss": 0.0,
      "distillation_loss": 0.3032427430152893,
      "epoch": 1.52,
      "learning_rate": 4.444908458338968e-05,
      "loss": 0.2889,
      "step": 1597,
      "task_loss": 0.16029399633407593
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7467726758252381,
      "compression_loss": 0.0,
      "distillation_loss": 0.22814524173736572,
      "epoch": 1.52,
      "learning_rate": 4.4442388023718624e-05,
      "loss": 0.231,
      "step": 1598,
      "task_loss": 0.25681257247924805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7467960621712019,
      "compression_loss": 0.0,
      "distillation_loss": 0.2185569703578949,
      "epoch": 1.52,
      "learning_rate": 4.443568793224415e-05,
      "loss": 0.221,
      "step": 1599,
      "task_loss": 0.24304306507110596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7468194416660316,
      "compression_loss": 0.0,
      "distillation_loss": 0.1327572613954544,
      "epoch": 1.52,
      "learning_rate": 4.4428984310183364e-05,
      "loss": 0.1352,
      "step": 1600,
      "task_loss": 0.15751878917217255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7468428143107311,
      "compression_loss": 0.0,
      "distillation_loss": 0.2773556709289551,
      "epoch": 1.52,
      "learning_rate": 4.4422277158754005e-05,
      "loss": 0.2637,
      "step": 1601,
      "task_loss": 0.14039871096611023
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.746866180106304,
      "compression_loss": 0.0,
      "distillation_loss": 0.2993885278701782,
      "epoch": 1.52,
      "learning_rate": 4.441556647917446e-05,
      "loss": 0.2958,
      "step": 1602,
      "task_loss": 0.26360023021698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.746889539053754,
      "compression_loss": 0.0,
      "distillation_loss": 0.14780689775943756,
      "epoch": 1.52,
      "learning_rate": 4.440885227266376e-05,
      "loss": 0.1404,
      "step": 1603,
      "task_loss": 0.07409326732158661
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7469128911540847,
      "compression_loss": 0.0,
      "distillation_loss": 0.1367529034614563,
      "epoch": 1.52,
      "learning_rate": 4.440213454044158e-05,
      "loss": 0.1366,
      "step": 1604,
      "task_loss": 0.13533815741539001
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7469362364082999,
      "compression_loss": 0.0,
      "distillation_loss": 0.11874578893184662,
      "epoch": 1.52,
      "learning_rate": 4.43954132837282e-05,
      "loss": 0.12,
      "step": 1605,
      "task_loss": 0.1314026564359665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7469595748174032,
      "compression_loss": 0.0,
      "distillation_loss": 0.09664209187030792,
      "epoch": 1.53,
      "learning_rate": 4.43886885037446e-05,
      "loss": 0.0971,
      "step": 1606,
      "task_loss": 0.10162815451622009
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7469829063823983,
      "compression_loss": 0.0,
      "distillation_loss": 0.08660341799259186,
      "epoch": 1.53,
      "learning_rate": 4.438196020171235e-05,
      "loss": 0.0801,
      "step": 1607,
      "task_loss": 0.021267052739858627
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7470062311042888,
      "compression_loss": 0.0,
      "distillation_loss": 0.08157442510128021,
      "epoch": 1.53,
      "learning_rate": 4.437522837885369e-05,
      "loss": 0.0841,
      "step": 1608,
      "task_loss": 0.10718289017677307
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7470295489840786,
      "compression_loss": 0.0,
      "distillation_loss": 0.13874168694019318,
      "epoch": 1.53,
      "learning_rate": 4.436849303639148e-05,
      "loss": 0.1323,
      "step": 1609,
      "task_loss": 0.07445013523101807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7470528600227712,
      "compression_loss": 0.0,
      "distillation_loss": 0.18086206912994385,
      "epoch": 1.53,
      "learning_rate": 4.436175417554923e-05,
      "loss": 0.1798,
      "step": 1610,
      "task_loss": 0.16998088359832764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7470761642213704,
      "compression_loss": 0.0,
      "distillation_loss": 0.11296135932207108,
      "epoch": 1.53,
      "learning_rate": 4.4355011797551086e-05,
      "loss": 0.1093,
      "step": 1611,
      "task_loss": 0.07632420212030411
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7470994615808797,
      "compression_loss": 0.0,
      "distillation_loss": 0.10066385567188263,
      "epoch": 1.53,
      "learning_rate": 4.4348265903621844e-05,
      "loss": 0.0935,
      "step": 1612,
      "task_loss": 0.028924619778990746
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.747122752102303,
      "compression_loss": 0.0,
      "distillation_loss": 0.0938473492860794,
      "epoch": 1.53,
      "learning_rate": 4.4341516494986904e-05,
      "loss": 0.107,
      "step": 1613,
      "task_loss": 0.22503921389579773
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7471460357866438,
      "compression_loss": 0.0,
      "distillation_loss": 0.06829311698675156,
      "epoch": 1.53,
      "learning_rate": 4.433476357287235e-05,
      "loss": 0.064,
      "step": 1614,
      "task_loss": 0.025221938267350197
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7471693126349059,
      "compression_loss": 0.0,
      "distillation_loss": 0.29679104685783386,
      "epoch": 1.53,
      "learning_rate": 4.432800713850488e-05,
      "loss": 0.3026,
      "step": 1615,
      "task_loss": 0.3548710346221924
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.747192582648093,
      "compression_loss": 0.0,
      "distillation_loss": 0.05282590910792351,
      "epoch": 1.53,
      "learning_rate": 4.432124719311182e-05,
      "loss": 0.0581,
      "step": 1616,
      "task_loss": 0.10553567111492157
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7472158458272087,
      "compression_loss": 0.0,
      "distillation_loss": 0.07523874193429947,
      "epoch": 1.54,
      "learning_rate": 4.431448373792116e-05,
      "loss": 0.0703,
      "step": 1617,
      "task_loss": 0.025965360924601555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7472391021732567,
      "compression_loss": 0.0,
      "distillation_loss": 0.24489791691303253,
      "epoch": 1.54,
      "learning_rate": 4.430771677416151e-05,
      "loss": 0.245,
      "step": 1618,
      "task_loss": 0.24620510637760162
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7472623516872408,
      "compression_loss": 0.0,
      "distillation_loss": 0.0306411050260067,
      "epoch": 1.54,
      "learning_rate": 4.430094630306212e-05,
      "loss": 0.0361,
      "step": 1619,
      "task_loss": 0.08524684607982635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7472855943701644,
      "compression_loss": 0.0,
      "distillation_loss": 0.1247522383928299,
      "epoch": 1.54,
      "learning_rate": 4.429417232585288e-05,
      "loss": 0.1176,
      "step": 1620,
      "task_loss": 0.05337625741958618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7473088302230314,
      "compression_loss": 0.0,
      "distillation_loss": 0.17402300238609314,
      "epoch": 1.54,
      "learning_rate": 4.428739484376431e-05,
      "loss": 0.1722,
      "step": 1621,
      "task_loss": 0.1557716578245163
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7473320592468454,
      "compression_loss": 0.0,
      "distillation_loss": 0.08097216486930847,
      "epoch": 1.54,
      "learning_rate": 4.4280613858027584e-05,
      "loss": 0.0769,
      "step": 1622,
      "task_loss": 0.039864055812358856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7473552814426102,
      "compression_loss": 0.0,
      "distillation_loss": 0.2641907334327698,
      "epoch": 1.54,
      "learning_rate": 4.427382936987449e-05,
      "loss": 0.2684,
      "step": 1623,
      "task_loss": 0.30657488107681274
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7473784968113295,
      "compression_loss": 0.0,
      "distillation_loss": 0.30244994163513184,
      "epoch": 1.54,
      "learning_rate": 4.426704138053747e-05,
      "loss": 0.2928,
      "step": 1624,
      "task_loss": 0.20630168914794922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7474017053540067,
      "compression_loss": 0.0,
      "distillation_loss": 0.08166351914405823,
      "epoch": 1.54,
      "learning_rate": 4.426024989124959e-05,
      "loss": 0.0787,
      "step": 1625,
      "task_loss": 0.05223226174712181
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7474249070716458,
      "compression_loss": 0.0,
      "distillation_loss": 0.044893529266119,
      "epoch": 1.54,
      "learning_rate": 4.425345490324456e-05,
      "loss": 0.052,
      "step": 1626,
      "task_loss": 0.11637747287750244
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7474481019652502,
      "compression_loss": 0.0,
      "distillation_loss": 0.17937017977237701,
      "epoch": 1.55,
      "learning_rate": 4.424665641775673e-05,
      "loss": 0.1818,
      "step": 1627,
      "task_loss": 0.20323118567466736
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7474712900358239,
      "compression_loss": 0.0,
      "distillation_loss": 0.24269092082977295,
      "epoch": 1.55,
      "learning_rate": 4.4239854436021056e-05,
      "loss": 0.2425,
      "step": 1628,
      "task_loss": 0.24110905826091766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7474944712843703,
      "compression_loss": 0.0,
      "distillation_loss": 0.13650964200496674,
      "epoch": 1.55,
      "learning_rate": 4.423304895927317e-05,
      "loss": 0.1435,
      "step": 1629,
      "task_loss": 0.20650048553943634
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7475176457118933,
      "compression_loss": 0.0,
      "distillation_loss": 0.15174457430839539,
      "epoch": 1.55,
      "learning_rate": 4.4226239988749305e-05,
      "loss": 0.1528,
      "step": 1630,
      "task_loss": 0.16226869821548462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7475408133193964,
      "compression_loss": 0.0,
      "distillation_loss": 0.17601847648620605,
      "epoch": 1.55,
      "learning_rate": 4.4219427525686366e-05,
      "loss": 0.1722,
      "step": 1631,
      "task_loss": 0.13755454123020172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7475639741078833,
      "compression_loss": 0.0,
      "distillation_loss": 0.09047359228134155,
      "epoch": 1.55,
      "learning_rate": 4.421261157132185e-05,
      "loss": 0.0868,
      "step": 1632,
      "task_loss": 0.053494758903980255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7475871280783579,
      "compression_loss": 0.0,
      "distillation_loss": 0.25244152545928955,
      "epoch": 1.55,
      "learning_rate": 4.4205792126893905e-05,
      "loss": 0.2515,
      "step": 1633,
      "task_loss": 0.24255844950675964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7476102752318236,
      "compression_loss": 0.0,
      "distillation_loss": 0.029148060828447342,
      "epoch": 1.55,
      "learning_rate": 4.4198969193641324e-05,
      "loss": 0.0272,
      "step": 1634,
      "task_loss": 0.009713640436530113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7476334155692843,
      "compression_loss": 0.0,
      "distillation_loss": 0.1328439861536026,
      "epoch": 1.55,
      "learning_rate": 4.4192142772803535e-05,
      "loss": 0.1342,
      "step": 1635,
      "task_loss": 0.14683926105499268
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7476565490917435,
      "compression_loss": 0.0,
      "distillation_loss": 0.20217649638652802,
      "epoch": 1.55,
      "learning_rate": 4.4185312865620575e-05,
      "loss": 0.192,
      "step": 1636,
      "task_loss": 0.1007172167301178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.747679675800205,
      "compression_loss": 0.0,
      "distillation_loss": 0.25275999307632446,
      "epoch": 1.55,
      "learning_rate": 4.417847947333314e-05,
      "loss": 0.2413,
      "step": 1637,
      "task_loss": 0.1383301168680191
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7477027956956724,
      "compression_loss": 0.0,
      "distillation_loss": 0.13600948452949524,
      "epoch": 1.56,
      "learning_rate": 4.417164259718254e-05,
      "loss": 0.1313,
      "step": 1638,
      "task_loss": 0.08905670791864395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7477259087791495,
      "compression_loss": 0.0,
      "distillation_loss": 0.24503761529922485,
      "epoch": 1.56,
      "learning_rate": 4.416480223841073e-05,
      "loss": 0.2425,
      "step": 1639,
      "task_loss": 0.21960334479808807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7477490150516399,
      "compression_loss": 0.0,
      "distillation_loss": 0.14646950364112854,
      "epoch": 1.56,
      "learning_rate": 4.4157958398260294e-05,
      "loss": 0.147,
      "step": 1640,
      "task_loss": 0.1517796665430069
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7477721145141473,
      "compression_loss": 0.0,
      "distillation_loss": 0.04808887839317322,
      "epoch": 1.56,
      "learning_rate": 4.415111107797445e-05,
      "loss": 0.0487,
      "step": 1641,
      "task_loss": 0.05420362949371338
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7477952071676754,
      "compression_loss": 0.0,
      "distillation_loss": 0.2355082482099533,
      "epoch": 1.56,
      "learning_rate": 4.414426027879705e-05,
      "loss": 0.2322,
      "step": 1642,
      "task_loss": 0.20212513208389282
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7478182930132279,
      "compression_loss": 0.0,
      "distillation_loss": 0.17253591120243073,
      "epoch": 1.56,
      "learning_rate": 4.413740600197257e-05,
      "loss": 0.1679,
      "step": 1643,
      "task_loss": 0.12654927372932434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7478413720518083,
      "compression_loss": 0.0,
      "distillation_loss": 0.30063819885253906,
      "epoch": 1.56,
      "learning_rate": 4.413054824874612e-05,
      "loss": 0.2869,
      "step": 1644,
      "task_loss": 0.16371384263038635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7478644442844206,
      "compression_loss": 0.0,
      "distillation_loss": 0.16720086336135864,
      "epoch": 1.56,
      "learning_rate": 4.412368702036345e-05,
      "loss": 0.1603,
      "step": 1645,
      "task_loss": 0.09819523990154266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7478875097120682,
      "compression_loss": 0.0,
      "distillation_loss": 0.12360204756259918,
      "epoch": 1.56,
      "learning_rate": 4.4116822318070925e-05,
      "loss": 0.116,
      "step": 1646,
      "task_loss": 0.047833651304244995
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.747910568335755,
      "compression_loss": 0.0,
      "distillation_loss": 0.08152472972869873,
      "epoch": 1.56,
      "learning_rate": 4.4109954143115565e-05,
      "loss": 0.0755,
      "step": 1647,
      "task_loss": 0.0212980005890131
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7479336201564845,
      "compression_loss": 0.0,
      "distillation_loss": 0.11956100910902023,
      "epoch": 1.57,
      "learning_rate": 4.4103082496745e-05,
      "loss": 0.1177,
      "step": 1648,
      "task_loss": 0.10143022984266281
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7479566651752604,
      "compression_loss": 0.0,
      "distillation_loss": 0.13289915025234222,
      "epoch": 1.57,
      "learning_rate": 4.40962073802075e-05,
      "loss": 0.1218,
      "step": 1649,
      "task_loss": 0.02191145159304142
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7479797033930865,
      "compression_loss": 0.0,
      "distillation_loss": 0.31535133719444275,
      "epoch": 1.57,
      "learning_rate": 4.4089328794751954e-05,
      "loss": 0.3183,
      "step": 1650,
      "task_loss": 0.3446260094642639
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7480027348109664,
      "compression_loss": 0.0,
      "distillation_loss": 0.04977913200855255,
      "epoch": 1.57,
      "learning_rate": 4.4082446741627906e-05,
      "loss": 0.0466,
      "step": 1651,
      "task_loss": 0.018024973571300507
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7480257594299038,
      "compression_loss": 0.0,
      "distillation_loss": 0.17800500988960266,
      "epoch": 1.57,
      "learning_rate": 4.40755612220855e-05,
      "loss": 0.1833,
      "step": 1652,
      "task_loss": 0.23083500564098358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7480487772509024,
      "compression_loss": 0.0,
      "distillation_loss": 0.07639691978693008,
      "epoch": 1.57,
      "learning_rate": 4.406867223737553e-05,
      "loss": 0.085,
      "step": 1653,
      "task_loss": 0.16263903677463531
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7480717882749659,
      "compression_loss": 0.0,
      "distillation_loss": 0.14029891788959503,
      "epoch": 1.57,
      "learning_rate": 4.406177978874941e-05,
      "loss": 0.1392,
      "step": 1654,
      "task_loss": 0.12951192259788513
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7480947925030978,
      "compression_loss": 0.0,
      "distillation_loss": 0.16938380897045135,
      "epoch": 1.57,
      "learning_rate": 4.405488387745919e-05,
      "loss": 0.1714,
      "step": 1655,
      "task_loss": 0.19002079963684082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7481177899363021,
      "compression_loss": 0.0,
      "distillation_loss": 0.1673574298620224,
      "epoch": 1.57,
      "learning_rate": 4.4047984504757544e-05,
      "loss": 0.1728,
      "step": 1656,
      "task_loss": 0.2220241129398346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7481407805755822,
      "compression_loss": 0.0,
      "distillation_loss": 0.1588287055492401,
      "epoch": 1.57,
      "learning_rate": 4.4041081671897775e-05,
      "loss": 0.15,
      "step": 1657,
      "task_loss": 0.07006968557834625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.748163764421942,
      "compression_loss": 0.0,
      "distillation_loss": 0.20809435844421387,
      "epoch": 1.57,
      "learning_rate": 4.403417538013382e-05,
      "loss": 0.2015,
      "step": 1658,
      "task_loss": 0.1424122452735901
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.748186741476385,
      "compression_loss": 0.0,
      "distillation_loss": 0.24815024435520172,
      "epoch": 1.58,
      "learning_rate": 4.402726563072024e-05,
      "loss": 0.2647,
      "step": 1659,
      "task_loss": 0.4135337769985199
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.748209711739915,
      "compression_loss": 0.0,
      "distillation_loss": 0.10159791260957718,
      "epoch": 1.58,
      "learning_rate": 4.4020352424912226e-05,
      "loss": 0.1108,
      "step": 1660,
      "task_loss": 0.193673774600029
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7482326752135356,
      "compression_loss": 0.0,
      "distillation_loss": 0.16284605860710144,
      "epoch": 1.58,
      "learning_rate": 4.401343576396558e-05,
      "loss": 0.1678,
      "step": 1661,
      "task_loss": 0.2123304009437561
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7482556318982505,
      "compression_loss": 0.0,
      "distillation_loss": 0.06474165618419647,
      "epoch": 1.58,
      "learning_rate": 4.400651564913676e-05,
      "loss": 0.0615,
      "step": 1662,
      "task_loss": 0.03229294717311859
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7482785817950635,
      "compression_loss": 0.0,
      "distillation_loss": 0.18344372510910034,
      "epoch": 1.58,
      "learning_rate": 4.399959208168284e-05,
      "loss": 0.1741,
      "step": 1663,
      "task_loss": 0.09026458114385605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7483015249049781,
      "compression_loss": 0.0,
      "distillation_loss": 0.19920691847801208,
      "epoch": 1.58,
      "learning_rate": 4.3992665062861514e-05,
      "loss": 0.1956,
      "step": 1664,
      "task_loss": 0.16354140639305115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7483244612289981,
      "compression_loss": 0.0,
      "distillation_loss": 0.09947105497121811,
      "epoch": 1.58,
      "learning_rate": 4.398573459393111e-05,
      "loss": 0.0958,
      "step": 1665,
      "task_loss": 0.06266459077596664
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7483473907681272,
      "compression_loss": 0.0,
      "distillation_loss": 0.12703344225883484,
      "epoch": 1.58,
      "learning_rate": 4.3978800676150575e-05,
      "loss": 0.1194,
      "step": 1666,
      "task_loss": 0.050371818244457245
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.748370313523369,
      "compression_loss": 0.0,
      "distillation_loss": 0.546389639377594,
      "epoch": 1.58,
      "learning_rate": 4.39718633107795e-05,
      "loss": 0.5258,
      "step": 1667,
      "task_loss": 0.3403877019882202
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7483932294957272,
      "compression_loss": 0.0,
      "distillation_loss": 0.0885571613907814,
      "epoch": 1.58,
      "learning_rate": 4.3964922499078084e-05,
      "loss": 0.0837,
      "step": 1668,
      "task_loss": 0.03993313014507294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7484161386862055,
      "compression_loss": 0.0,
      "distillation_loss": 0.1787647306919098,
      "epoch": 1.58,
      "learning_rate": 4.3957978242307166e-05,
      "loss": 0.175,
      "step": 1669,
      "task_loss": 0.14156457781791687
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7484390410958076,
      "compression_loss": 0.0,
      "distillation_loss": 0.11095503717660904,
      "epoch": 1.59,
      "learning_rate": 4.395103054172819e-05,
      "loss": 0.1135,
      "step": 1670,
      "task_loss": 0.13681727647781372
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7484619367255372,
      "compression_loss": 0.0,
      "distillation_loss": 0.09957759082317352,
      "epoch": 1.59,
      "learning_rate": 4.394407939860325e-05,
      "loss": 0.1075,
      "step": 1671,
      "task_loss": 0.17913945019245148
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7484848255763978,
      "compression_loss": 0.0,
      "distillation_loss": 0.14932695031166077,
      "epoch": 1.59,
      "learning_rate": 4.3937124814195054e-05,
      "loss": 0.1552,
      "step": 1672,
      "task_loss": 0.20759890973567963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7485077076493933,
      "compression_loss": 0.0,
      "distillation_loss": 0.11619941145181656,
      "epoch": 1.59,
      "learning_rate": 4.393016678976692e-05,
      "loss": 0.1115,
      "step": 1673,
      "task_loss": 0.06885246187448502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7485305829455273,
      "compression_loss": 0.0,
      "distillation_loss": 0.12013675272464752,
      "epoch": 1.59,
      "learning_rate": 4.3923205326582837e-05,
      "loss": 0.1239,
      "step": 1674,
      "task_loss": 0.15799427032470703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7485534514658035,
      "compression_loss": 0.0,
      "distillation_loss": 0.08647421002388,
      "epoch": 1.59,
      "learning_rate": 4.3916240425907364e-05,
      "loss": 0.0989,
      "step": 1675,
      "task_loss": 0.2103899121284485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7485763132112256,
      "compression_loss": 0.0,
      "distillation_loss": 0.25060856342315674,
      "epoch": 1.59,
      "learning_rate": 4.3909272089005714e-05,
      "loss": 0.2436,
      "step": 1676,
      "task_loss": 0.18063423037528992
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7485991681827973,
      "compression_loss": 0.0,
      "distillation_loss": 0.23684711754322052,
      "epoch": 1.59,
      "learning_rate": 4.3902300317143726e-05,
      "loss": 0.2305,
      "step": 1677,
      "task_loss": 0.17349837720394135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7486220163815221,
      "compression_loss": 0.0,
      "distillation_loss": 0.2393345683813095,
      "epoch": 1.59,
      "learning_rate": 4.389532511158785e-05,
      "loss": 0.2251,
      "step": 1678,
      "task_loss": 0.09700442850589752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.748644857808404,
      "compression_loss": 0.0,
      "distillation_loss": 0.1394980251789093,
      "epoch": 1.59,
      "learning_rate": 4.388834647360516e-05,
      "loss": 0.1531,
      "step": 1679,
      "task_loss": 0.2758055329322815
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7486676924644463,
      "compression_loss": 0.0,
      "distillation_loss": 0.08193551748991013,
      "epoch": 1.6,
      "learning_rate": 4.388136440446337e-05,
      "loss": 0.0893,
      "step": 1680,
      "task_loss": 0.15590821206569672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.748690520350653,
      "compression_loss": 0.0,
      "distillation_loss": 0.032735083252191544,
      "epoch": 1.6,
      "learning_rate": 4.387437890543081e-05,
      "loss": 0.0302,
      "step": 1681,
      "task_loss": 0.007712380960583687
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7487133414680276,
      "compression_loss": 0.0,
      "distillation_loss": 0.2094217985868454,
      "epoch": 1.6,
      "learning_rate": 4.3867389977776416e-05,
      "loss": 0.2026,
      "step": 1682,
      "task_loss": 0.14127467572689056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7487361558175739,
      "compression_loss": 0.0,
      "distillation_loss": 0.14119529724121094,
      "epoch": 1.6,
      "learning_rate": 4.3860397622769756e-05,
      "loss": 0.1352,
      "step": 1683,
      "task_loss": 0.08111706376075745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7487589634002955,
      "compression_loss": 0.0,
      "distillation_loss": 0.2153528779745102,
      "epoch": 1.6,
      "learning_rate": 4.3853401841681046e-05,
      "loss": 0.2241,
      "step": 1684,
      "task_loss": 0.3023759126663208
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7487817642171961,
      "compression_loss": 0.0,
      "distillation_loss": 0.050505850464105606,
      "epoch": 1.6,
      "learning_rate": 4.3846402635781093e-05,
      "loss": 0.0562,
      "step": 1685,
      "task_loss": 0.1069660410284996
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7488045582692795,
      "compression_loss": 0.0,
      "distillation_loss": 0.1630827784538269,
      "epoch": 1.6,
      "learning_rate": 4.3839400006341335e-05,
      "loss": 0.1598,
      "step": 1686,
      "task_loss": 0.1302591860294342
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7488273455575492,
      "compression_loss": 0.0,
      "distillation_loss": 0.03398447483778,
      "epoch": 1.6,
      "learning_rate": 4.383239395463383e-05,
      "loss": 0.0397,
      "step": 1687,
      "task_loss": 0.090923011302948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7488501260830089,
      "compression_loss": 0.0,
      "distillation_loss": 0.06578654050827026,
      "epoch": 1.6,
      "learning_rate": 4.382538448193127e-05,
      "loss": 0.0629,
      "step": 1688,
      "task_loss": 0.037396807223558426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7488728998466624,
      "compression_loss": 0.0,
      "distillation_loss": 0.11456318199634552,
      "epoch": 1.6,
      "learning_rate": 4.381837158950695e-05,
      "loss": 0.1149,
      "step": 1689,
      "task_loss": 0.11771449446678162
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7488956668495133,
      "compression_loss": 0.0,
      "distillation_loss": 0.09141987562179565,
      "epoch": 1.6,
      "learning_rate": 4.3811355278634804e-05,
      "loss": 0.0861,
      "step": 1690,
      "task_loss": 0.03799279406666756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7489184270925653,
      "compression_loss": 0.0,
      "distillation_loss": 0.07688391953706741,
      "epoch": 1.61,
      "learning_rate": 4.380433555058937e-05,
      "loss": 0.0735,
      "step": 1691,
      "task_loss": 0.04309658333659172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7489411805768221,
      "compression_loss": 0.0,
      "distillation_loss": 0.24183808267116547,
      "epoch": 1.61,
      "learning_rate": 4.379731240664583e-05,
      "loss": 0.2428,
      "step": 1692,
      "task_loss": 0.251809298992157
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7489639273032873,
      "compression_loss": 0.0,
      "distillation_loss": 0.04763605445623398,
      "epoch": 1.61,
      "learning_rate": 4.379028584807996e-05,
      "loss": 0.0519,
      "step": 1693,
      "task_loss": 0.09006030112504959
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7489866672729648,
      "compression_loss": 0.0,
      "distillation_loss": 0.26702186465263367,
      "epoch": 1.61,
      "learning_rate": 4.3783255876168165e-05,
      "loss": 0.2575,
      "step": 1694,
      "task_loss": 0.17218661308288574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.749009400486858,
      "compression_loss": 0.0,
      "distillation_loss": 0.1826254427433014,
      "epoch": 1.61,
      "learning_rate": 4.377622249218748e-05,
      "loss": 0.1761,
      "step": 1695,
      "task_loss": 0.11716414242982864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7490321269459708,
      "compression_loss": 0.0,
      "distillation_loss": 0.06596788763999939,
      "epoch": 1.61,
      "learning_rate": 4.376918569741556e-05,
      "loss": 0.078,
      "step": 1696,
      "task_loss": 0.18661688268184662
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7490548466513067,
      "compression_loss": 0.0,
      "distillation_loss": 0.07024285942316055,
      "epoch": 1.61,
      "learning_rate": 4.376214549313066e-05,
      "loss": 0.0653,
      "step": 1697,
      "task_loss": 0.020744740962982178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7490775596038696,
      "compression_loss": 0.0,
      "distillation_loss": 0.13795125484466553,
      "epoch": 1.61,
      "learning_rate": 4.375510188061167e-05,
      "loss": 0.1382,
      "step": 1698,
      "task_loss": 0.14067493379116058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.749100265804663,
      "compression_loss": 0.0,
      "distillation_loss": 0.06560136377811432,
      "epoch": 1.61,
      "learning_rate": 4.37480548611381e-05,
      "loss": 0.0745,
      "step": 1699,
      "task_loss": 0.15437015891075134
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7491229652546907,
      "compression_loss": 0.0,
      "distillation_loss": 0.10080569982528687,
      "epoch": 1.61,
      "learning_rate": 4.374100443599007e-05,
      "loss": 0.0956,
      "step": 1700,
      "task_loss": 0.04915327578783035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7491456579549562,
      "compression_loss": 0.0,
      "distillation_loss": 0.151437908411026,
      "epoch": 1.62,
      "learning_rate": 4.3733950606448324e-05,
      "loss": 0.1444,
      "step": 1701,
      "task_loss": 0.08063948154449463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7491683439064634,
      "compression_loss": 0.0,
      "distillation_loss": 0.14637860655784607,
      "epoch": 1.62,
      "learning_rate": 4.3726893373794234e-05,
      "loss": 0.1415,
      "step": 1702,
      "task_loss": 0.09764857590198517
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7491910231102158,
      "compression_loss": 0.0,
      "distillation_loss": 0.18097403645515442,
      "epoch": 1.62,
      "learning_rate": 4.3719832739309766e-05,
      "loss": 0.1771,
      "step": 1703,
      "task_loss": 0.1424616128206253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7492136955672173,
      "compression_loss": 0.0,
      "distillation_loss": 0.22387400269508362,
      "epoch": 1.62,
      "learning_rate": 4.371276870427753e-05,
      "loss": 0.2137,
      "step": 1704,
      "task_loss": 0.12188294529914856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7492363612784714,
      "compression_loss": 0.0,
      "distillation_loss": 0.09288572520017624,
      "epoch": 1.62,
      "learning_rate": 4.3705701269980734e-05,
      "loss": 0.0888,
      "step": 1705,
      "task_loss": 0.05178219825029373
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7492590202449818,
      "compression_loss": 0.0,
      "distillation_loss": 0.05206472426652908,
      "epoch": 1.62,
      "learning_rate": 4.369863043770322e-05,
      "loss": 0.048,
      "step": 1706,
      "task_loss": 0.01116347685456276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7492816724677522,
      "compression_loss": 0.0,
      "distillation_loss": 0.11153702437877655,
      "epoch": 1.62,
      "learning_rate": 4.369155620872943e-05,
      "loss": 0.1047,
      "step": 1707,
      "task_loss": 0.04338126629590988
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7493043179477863,
      "compression_loss": 0.0,
      "distillation_loss": 0.15479451417922974,
      "epoch": 1.62,
      "learning_rate": 4.3684478584344433e-05,
      "loss": 0.1515,
      "step": 1708,
      "task_loss": 0.12206155061721802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7493269566860878,
      "compression_loss": 0.0,
      "distillation_loss": 0.05233832448720932,
      "epoch": 1.62,
      "learning_rate": 4.367739756583392e-05,
      "loss": 0.0498,
      "step": 1709,
      "task_loss": 0.026678021997213364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7493495886836604,
      "compression_loss": 0.0,
      "distillation_loss": 0.05933618173003197,
      "epoch": 1.62,
      "learning_rate": 4.367031315448419e-05,
      "loss": 0.0733,
      "step": 1710,
      "task_loss": 0.19880297780036926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7493722139415077,
      "compression_loss": 0.0,
      "distillation_loss": 0.1785992830991745,
      "epoch": 1.62,
      "learning_rate": 4.366322535158215e-05,
      "loss": 0.1674,
      "step": 1711,
      "task_loss": 0.06677613407373428
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7493948324606334,
      "compression_loss": 0.0,
      "distillation_loss": 0.05169357359409332,
      "epoch": 1.63,
      "learning_rate": 4.3656134158415344e-05,
      "loss": 0.0492,
      "step": 1712,
      "task_loss": 0.026786495000123978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7494174442420414,
      "compression_loss": 0.0,
      "distillation_loss": 0.09955106675624847,
      "epoch": 1.63,
      "learning_rate": 4.364903957627192e-05,
      "loss": 0.0959,
      "step": 1713,
      "task_loss": 0.06332787871360779
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.749440049286735,
      "compression_loss": 0.0,
      "distillation_loss": 0.20070070028305054,
      "epoch": 1.63,
      "learning_rate": 4.3641941606440644e-05,
      "loss": 0.1948,
      "step": 1714,
      "task_loss": 0.14132705330848694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7494626475957181,
      "compression_loss": 0.0,
      "distillation_loss": 0.07433759421110153,
      "epoch": 1.63,
      "learning_rate": 4.36348402502109e-05,
      "loss": 0.0819,
      "step": 1715,
      "task_loss": 0.15012231469154358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7494852391699944,
      "compression_loss": 0.0,
      "distillation_loss": 0.13596729934215546,
      "epoch": 1.63,
      "learning_rate": 4.3627735508872666e-05,
      "loss": 0.1536,
      "step": 1716,
      "task_loss": 0.31197068095207214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7495078240105675,
      "compression_loss": 0.0,
      "distillation_loss": 0.2269076108932495,
      "epoch": 1.63,
      "learning_rate": 4.362062738371657e-05,
      "loss": 0.2384,
      "step": 1717,
      "task_loss": 0.3419533371925354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7495304021184411,
      "compression_loss": 0.0,
      "distillation_loss": 0.20675890147686005,
      "epoch": 1.63,
      "learning_rate": 4.361351587603384e-05,
      "loss": 0.2075,
      "step": 1718,
      "task_loss": 0.21458357572555542
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7495529734946189,
      "compression_loss": 0.0,
      "distillation_loss": 0.10246886312961578,
      "epoch": 1.63,
      "learning_rate": 4.360640098711629e-05,
      "loss": 0.0964,
      "step": 1719,
      "task_loss": 0.04224860668182373
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7495755381401047,
      "compression_loss": 0.0,
      "distillation_loss": 0.250265896320343,
      "epoch": 1.63,
      "learning_rate": 4.3599282718256406e-05,
      "loss": 0.2593,
      "step": 1720,
      "task_loss": 0.34033694863319397
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.749598096055902,
      "compression_loss": 0.0,
      "distillation_loss": 0.11491803824901581,
      "epoch": 1.63,
      "learning_rate": 4.3592161070747233e-05,
      "loss": 0.1297,
      "step": 1721,
      "task_loss": 0.2628564238548279
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7496206472430145,
      "compression_loss": 0.0,
      "distillation_loss": 0.05386704206466675,
      "epoch": 1.64,
      "learning_rate": 4.358503604588247e-05,
      "loss": 0.0536,
      "step": 1722,
      "task_loss": 0.050750650465488434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.749643191702446,
      "compression_loss": 0.0,
      "distillation_loss": 0.35980457067489624,
      "epoch": 1.64,
      "learning_rate": 4.357790764495639e-05,
      "loss": 0.3559,
      "step": 1723,
      "task_loss": 0.3210410475730896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7496657294352,
      "compression_loss": 0.0,
      "distillation_loss": 0.08513958752155304,
      "epoch": 1.64,
      "learning_rate": 4.357077586926392e-05,
      "loss": 0.0913,
      "step": 1724,
      "task_loss": 0.14632166922092438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7496882604422804,
      "compression_loss": 0.0,
      "distillation_loss": 0.16130788624286652,
      "epoch": 1.64,
      "learning_rate": 4.356364072010059e-05,
      "loss": 0.1687,
      "step": 1725,
      "task_loss": 0.23540650308132172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7497107847246908,
      "compression_loss": 0.0,
      "distillation_loss": 0.10402365028858185,
      "epoch": 1.64,
      "learning_rate": 4.3556502198762496e-05,
      "loss": 0.1023,
      "step": 1726,
      "task_loss": 0.08679551631212234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7497333022834348,
      "compression_loss": 0.0,
      "distillation_loss": 0.07432233542203903,
      "epoch": 1.64,
      "learning_rate": 4.354936030654642e-05,
      "loss": 0.0807,
      "step": 1727,
      "task_loss": 0.13773983716964722
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7497558131195161,
      "compression_loss": 0.0,
      "distillation_loss": 0.2607779800891876,
      "epoch": 1.64,
      "learning_rate": 4.3542215044749705e-05,
      "loss": 0.2518,
      "step": 1728,
      "task_loss": 0.1711687445640564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7497783172339385,
      "compression_loss": 0.0,
      "distillation_loss": 0.059374839067459106,
      "epoch": 1.64,
      "learning_rate": 4.3535066414670336e-05,
      "loss": 0.0628,
      "step": 1729,
      "task_loss": 0.09339209645986557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7498008146277056,
      "compression_loss": 0.0,
      "distillation_loss": 0.20149439573287964,
      "epoch": 1.64,
      "learning_rate": 4.352791441760687e-05,
      "loss": 0.1973,
      "step": 1730,
      "task_loss": 0.15943476557731628
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7498233053018211,
      "compression_loss": 0.0,
      "distillation_loss": 0.13185396790504456,
      "epoch": 1.64,
      "learning_rate": 4.352075905485854e-05,
      "loss": 0.1234,
      "step": 1731,
      "task_loss": 0.04712344706058502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7498457892572886,
      "compression_loss": 0.0,
      "distillation_loss": 0.1501794159412384,
      "epoch": 1.64,
      "learning_rate": 4.351360032772512e-05,
      "loss": 0.1726,
      "step": 1732,
      "task_loss": 0.3744635581970215
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.749868266495112,
      "compression_loss": 0.0,
      "distillation_loss": 0.07114310562610626,
      "epoch": 1.65,
      "learning_rate": 4.3506438237507033e-05,
      "loss": 0.0689,
      "step": 1733,
      "task_loss": 0.048517607152462006
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7498907370162946,
      "compression_loss": 0.0,
      "distillation_loss": 0.13893577456474304,
      "epoch": 1.65,
      "learning_rate": 4.3499272785505316e-05,
      "loss": 0.159,
      "step": 1734,
      "task_loss": 0.33932289481163025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7499132008218404,
      "compression_loss": 0.0,
      "distillation_loss": 0.14102880656719208,
      "epoch": 1.65,
      "learning_rate": 4.349210397302161e-05,
      "loss": 0.1395,
      "step": 1735,
      "task_loss": 0.12529143691062927
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7499356579127531,
      "compression_loss": 0.0,
      "distillation_loss": 0.2732577919960022,
      "epoch": 1.65,
      "learning_rate": 4.348493180135815e-05,
      "loss": 0.2615,
      "step": 1736,
      "task_loss": 0.15519018471240997
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7499581082900362,
      "compression_loss": 0.0,
      "distillation_loss": 0.061276476830244064,
      "epoch": 1.65,
      "learning_rate": 4.347775627181782e-05,
      "loss": 0.0631,
      "step": 1737,
      "task_loss": 0.07924476265907288
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7499805519546935,
      "compression_loss": 0.0,
      "distillation_loss": 0.20356178283691406,
      "epoch": 1.65,
      "learning_rate": 4.3470577385704056e-05,
      "loss": 0.2137,
      "step": 1738,
      "task_loss": 0.3052142560482025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7500029889077287,
      "compression_loss": 0.0,
      "distillation_loss": 0.24523907899856567,
      "epoch": 1.65,
      "learning_rate": 4.346339514432096e-05,
      "loss": 0.2379,
      "step": 1739,
      "task_loss": 0.17217698693275452
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7500254191501452,
      "compression_loss": 0.0,
      "distillation_loss": 0.14814287424087524,
      "epoch": 1.65,
      "learning_rate": 4.345620954897322e-05,
      "loss": 0.1381,
      "step": 1740,
      "task_loss": 0.04762396588921547
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7500478426829471,
      "compression_loss": 0.0,
      "distillation_loss": 0.08230330049991608,
      "epoch": 1.65,
      "learning_rate": 4.344902060096612e-05,
      "loss": 0.0829,
      "step": 1741,
      "task_loss": 0.08790513873100281
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7500702595071379,
      "compression_loss": 0.0,
      "distillation_loss": 0.12028224766254425,
      "epoch": 1.65,
      "learning_rate": 4.344182830160558e-05,
      "loss": 0.1143,
      "step": 1742,
      "task_loss": 0.060103029012680054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7500926696237211,
      "compression_loss": 0.0,
      "distillation_loss": 0.12078093737363815,
      "epoch": 1.66,
      "learning_rate": 4.343463265219811e-05,
      "loss": 0.115,
      "step": 1743,
      "task_loss": 0.06334509700536728
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7501150730337007,
      "compression_loss": 0.0,
      "distillation_loss": 0.10595209896564484,
      "epoch": 1.66,
      "learning_rate": 4.342743365405084e-05,
      "loss": 0.1103,
      "step": 1744,
      "task_loss": 0.14954574406147003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7501374697380803,
      "compression_loss": 0.0,
      "distillation_loss": 0.16210979223251343,
      "epoch": 1.66,
      "learning_rate": 4.3420231308471496e-05,
      "loss": 0.1555,
      "step": 1745,
      "task_loss": 0.09629593789577484
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7501598597378634,
      "compression_loss": 0.0,
      "distillation_loss": 0.185985267162323,
      "epoch": 1.66,
      "learning_rate": 4.3413025616768424e-05,
      "loss": 0.1858,
      "step": 1746,
      "task_loss": 0.1842183768749237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7501822430340538,
      "compression_loss": 0.0,
      "distillation_loss": 0.1382254958152771,
      "epoch": 1.66,
      "learning_rate": 4.340581658025058e-05,
      "loss": 0.1359,
      "step": 1747,
      "task_loss": 0.11496403813362122
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7502046196276552,
      "compression_loss": 0.0,
      "distillation_loss": 0.0522196963429451,
      "epoch": 1.66,
      "learning_rate": 4.33986042002275e-05,
      "loss": 0.058,
      "step": 1748,
      "task_loss": 0.10959599912166595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7502269895196713,
      "compression_loss": 0.0,
      "distillation_loss": 0.045316457748413086,
      "epoch": 1.66,
      "learning_rate": 4.339138847800936e-05,
      "loss": 0.0488,
      "step": 1749,
      "task_loss": 0.08048881590366364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7502493527111057,
      "compression_loss": 0.0,
      "distillation_loss": 0.11536607146263123,
      "epoch": 1.66,
      "learning_rate": 4.3384169414906925e-05,
      "loss": 0.1198,
      "step": 1750,
      "task_loss": 0.1599726378917694
    },
    {
      "epoch": 1.66,
      "eval_accuracy": 0.9002293577981652,
      "eval_loss": 0.3980746567249298,
      "eval_runtime": 18.3137,
      "eval_samples_per_second": 47.615,
      "eval_steps_per_second": 5.952,
      "step": 1750
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7502717092029622,
      "compression_loss": 0.0,
      "distillation_loss": 0.22586020827293396,
      "epoch": 1.66,
      "learning_rate": 4.3376947012231586e-05,
      "loss": 0.214,
      "step": 1751,
      "task_loss": 0.10768207907676697
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7502940589962444,
      "compression_loss": 0.0,
      "distillation_loss": 0.10226061195135117,
      "epoch": 1.66,
      "learning_rate": 4.336972127129532e-05,
      "loss": 0.1036,
      "step": 1752,
      "task_loss": 0.11605414748191833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7503164020919558,
      "compression_loss": 0.0,
      "distillation_loss": 0.2922346591949463,
      "epoch": 1.66,
      "learning_rate": 4.3362492193410705e-05,
      "loss": 0.2796,
      "step": 1753,
      "task_loss": 0.1663048416376114
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7503387384911004,
      "compression_loss": 0.0,
      "distillation_loss": 0.05565609037876129,
      "epoch": 1.67,
      "learning_rate": 4.335525977989095e-05,
      "loss": 0.0522,
      "step": 1754,
      "task_loss": 0.020734498277306557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7503610681946818,
      "compression_loss": 0.0,
      "distillation_loss": 0.19371697306632996,
      "epoch": 1.67,
      "learning_rate": 4.334802403204986e-05,
      "loss": 0.1989,
      "step": 1755,
      "task_loss": 0.2459201216697693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7503833912037037,
      "compression_loss": 0.0,
      "distillation_loss": 0.12561503052711487,
      "epoch": 1.67,
      "learning_rate": 4.334078495120184e-05,
      "loss": 0.1226,
      "step": 1756,
      "task_loss": 0.0955725908279419
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7504057075191697,
      "compression_loss": 0.0,
      "distillation_loss": 0.25841909646987915,
      "epoch": 1.67,
      "learning_rate": 4.33335425386619e-05,
      "loss": 0.2526,
      "step": 1757,
      "task_loss": 0.2000465989112854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7504280171420834,
      "compression_loss": 0.0,
      "distillation_loss": 0.18024028837680817,
      "epoch": 1.67,
      "learning_rate": 4.332629679574566e-05,
      "loss": 0.1706,
      "step": 1758,
      "task_loss": 0.08373329043388367
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7504503200734487,
      "compression_loss": 0.0,
      "distillation_loss": 0.21449556946754456,
      "epoch": 1.67,
      "learning_rate": 4.331904772376935e-05,
      "loss": 0.2073,
      "step": 1759,
      "task_loss": 0.14211583137512207
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7504726163142691,
      "compression_loss": 0.0,
      "distillation_loss": 0.17756842076778412,
      "epoch": 1.67,
      "learning_rate": 4.3311795324049795e-05,
      "loss": 0.1765,
      "step": 1760,
      "task_loss": 0.16666541993618011
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7504949058655483,
      "compression_loss": 0.0,
      "distillation_loss": 0.1488828957080841,
      "epoch": 1.67,
      "learning_rate": 4.3304539597904435e-05,
      "loss": 0.1445,
      "step": 1761,
      "task_loss": 0.10537352412939072
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7505171887282901,
      "compression_loss": 0.0,
      "distillation_loss": 0.18257030844688416,
      "epoch": 1.67,
      "learning_rate": 4.3297280546651295e-05,
      "loss": 0.185,
      "step": 1762,
      "task_loss": 0.20700550079345703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7505394649034981,
      "compression_loss": 0.0,
      "distillation_loss": 0.04670891910791397,
      "epoch": 1.67,
      "learning_rate": 4.329001817160903e-05,
      "loss": 0.05,
      "step": 1763,
      "task_loss": 0.07921046018600464
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.750561734392176,
      "compression_loss": 0.0,
      "distillation_loss": 0.15156331658363342,
      "epoch": 1.68,
      "learning_rate": 4.3282752474096864e-05,
      "loss": 0.1615,
      "step": 1764,
      "task_loss": 0.2507190704345703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7505839971953274,
      "compression_loss": 0.0,
      "distillation_loss": 0.23599442839622498,
      "epoch": 1.68,
      "learning_rate": 4.327548345543467e-05,
      "loss": 0.2256,
      "step": 1765,
      "task_loss": 0.13191649317741394
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7506062533139561,
      "compression_loss": 0.0,
      "distillation_loss": 0.15651676058769226,
      "epoch": 1.68,
      "learning_rate": 4.326821111694289e-05,
      "loss": 0.1509,
      "step": 1766,
      "task_loss": 0.10070617496967316
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7506285027490659,
      "compression_loss": 0.0,
      "distillation_loss": 0.14649711549282074,
      "epoch": 1.68,
      "learning_rate": 4.3260935459942584e-05,
      "loss": 0.144,
      "step": 1767,
      "task_loss": 0.12118849158287048
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7506507455016602,
      "compression_loss": 0.0,
      "distillation_loss": 0.05859662592411041,
      "epoch": 1.68,
      "learning_rate": 4.32536564857554e-05,
      "loss": 0.0547,
      "step": 1768,
      "task_loss": 0.020115777850151062
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7506729815727428,
      "compression_loss": 0.0,
      "distillation_loss": 0.24223226308822632,
      "epoch": 1.68,
      "learning_rate": 4.3246374195703604e-05,
      "loss": 0.2509,
      "step": 1769,
      "task_loss": 0.32900571823120117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7506952109633174,
      "compression_loss": 0.0,
      "distillation_loss": 0.11694598197937012,
      "epoch": 1.68,
      "learning_rate": 4.3239088591110065e-05,
      "loss": 0.1294,
      "step": 1770,
      "task_loss": 0.2414626181125641
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7507174336743877,
      "compression_loss": 0.0,
      "distillation_loss": 0.23970989882946014,
      "epoch": 1.68,
      "learning_rate": 4.323179967329824e-05,
      "loss": 0.2291,
      "step": 1771,
      "task_loss": 0.13375961780548096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7507396497069573,
      "compression_loss": 0.0,
      "distillation_loss": 0.050326522439718246,
      "epoch": 1.68,
      "learning_rate": 4.3224507443592196e-05,
      "loss": 0.0593,
      "step": 1772,
      "task_loss": 0.13998878002166748
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.75076185906203,
      "compression_loss": 0.0,
      "distillation_loss": 0.23953363299369812,
      "epoch": 1.68,
      "learning_rate": 4.321721190331661e-05,
      "loss": 0.2254,
      "step": 1773,
      "task_loss": 0.09803837537765503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7507840617406093,
      "compression_loss": 0.0,
      "distillation_loss": 0.12072421610355377,
      "epoch": 1.68,
      "learning_rate": 4.3209913053796746e-05,
      "loss": 0.1307,
      "step": 1774,
      "task_loss": 0.2204258143901825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7508062577436991,
      "compression_loss": 0.0,
      "distillation_loss": 0.2765160799026489,
      "epoch": 1.69,
      "learning_rate": 4.3202610896358474e-05,
      "loss": 0.2627,
      "step": 1775,
      "task_loss": 0.13829368352890015
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7508284470723029,
      "compression_loss": 0.0,
      "distillation_loss": 0.16436317563056946,
      "epoch": 1.69,
      "learning_rate": 4.319530543232827e-05,
      "loss": 0.1646,
      "step": 1776,
      "task_loss": 0.16645075380802155
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7508506297274246,
      "compression_loss": 0.0,
      "distillation_loss": 0.1174478605389595,
      "epoch": 1.69,
      "learning_rate": 4.31879966630332e-05,
      "loss": 0.1172,
      "step": 1777,
      "task_loss": 0.11536243557929993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7508728057100676,
      "compression_loss": 0.0,
      "distillation_loss": 0.33932632207870483,
      "epoch": 1.69,
      "learning_rate": 4.318068458980095e-05,
      "loss": 0.3292,
      "step": 1778,
      "task_loss": 0.23805338144302368
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7508949750212358,
      "compression_loss": 0.0,
      "distillation_loss": 0.08245585113763809,
      "epoch": 1.69,
      "learning_rate": 4.317336921395978e-05,
      "loss": 0.0889,
      "step": 1779,
      "task_loss": 0.14642596244812012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7509171376619328,
      "compression_loss": 0.0,
      "distillation_loss": 0.32606542110443115,
      "epoch": 1.69,
      "learning_rate": 4.316605053683856e-05,
      "loss": 0.3143,
      "step": 1780,
      "task_loss": 0.20817196369171143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7509392936331623,
      "compression_loss": 0.0,
      "distillation_loss": 0.10882005095481873,
      "epoch": 1.69,
      "learning_rate": 4.3158728559766786e-05,
      "loss": 0.1177,
      "step": 1781,
      "task_loss": 0.1978674679994583
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7509614429359279,
      "compression_loss": 0.0,
      "distillation_loss": 0.0932815819978714,
      "epoch": 1.69,
      "learning_rate": 4.315140328407451e-05,
      "loss": 0.0889,
      "step": 1782,
      "task_loss": 0.0497298426926136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7509835855712335,
      "compression_loss": 0.0,
      "distillation_loss": 0.12256644666194916,
      "epoch": 1.69,
      "learning_rate": 4.314407471109241e-05,
      "loss": 0.1198,
      "step": 1783,
      "task_loss": 0.0945819541811943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7510057215400826,
      "compression_loss": 0.0,
      "distillation_loss": 0.09638384729623795,
      "epoch": 1.69,
      "learning_rate": 4.313674284215176e-05,
      "loss": 0.0956,
      "step": 1784,
      "task_loss": 0.08847030997276306
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7510278508434789,
      "compression_loss": 0.0,
      "distillation_loss": 0.15519554913043976,
      "epoch": 1.7,
      "learning_rate": 4.312940767858441e-05,
      "loss": 0.1552,
      "step": 1785,
      "task_loss": 0.15559692680835724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7510499734824261,
      "compression_loss": 0.0,
      "distillation_loss": 0.09053555130958557,
      "epoch": 1.7,
      "learning_rate": 4.312206922172286e-05,
      "loss": 0.0836,
      "step": 1786,
      "task_loss": 0.021591845899820328
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7510720894579278,
      "compression_loss": 0.0,
      "distillation_loss": 0.0775122344493866,
      "epoch": 1.7,
      "learning_rate": 4.311472747290015e-05,
      "loss": 0.0729,
      "step": 1787,
      "task_loss": 0.03092704340815544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7510941987709878,
      "compression_loss": 0.0,
      "distillation_loss": 0.20807501673698425,
      "epoch": 1.7,
      "learning_rate": 4.310738243344996e-05,
      "loss": 0.2011,
      "step": 1788,
      "task_loss": 0.1382240504026413
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7511163014226098,
      "compression_loss": 0.0,
      "distillation_loss": 0.13786278665065765,
      "epoch": 1.7,
      "learning_rate": 4.310003410470653e-05,
      "loss": 0.137,
      "step": 1789,
      "task_loss": 0.12924642860889435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7511383974137974,
      "compression_loss": 0.0,
      "distillation_loss": 0.03874251991510391,
      "epoch": 1.7,
      "learning_rate": 4.309268248800476e-05,
      "loss": 0.0455,
      "step": 1790,
      "task_loss": 0.10590145736932755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7511604867455544,
      "compression_loss": 0.0,
      "distillation_loss": 0.16585403680801392,
      "epoch": 1.7,
      "learning_rate": 4.3085327584680056e-05,
      "loss": 0.1747,
      "step": 1791,
      "task_loss": 0.2540725767612457
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7511825694188843,
      "compression_loss": 0.0,
      "distillation_loss": 0.17077066004276276,
      "epoch": 1.7,
      "learning_rate": 4.3077969396068505e-05,
      "loss": 0.172,
      "step": 1792,
      "task_loss": 0.18328894674777985
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7512046454347909,
      "compression_loss": 0.0,
      "distillation_loss": 0.23324266076087952,
      "epoch": 1.7,
      "learning_rate": 4.307060792350675e-05,
      "loss": 0.2235,
      "step": 1793,
      "task_loss": 0.13554205000400543
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7512267147942778,
      "compression_loss": 0.0,
      "distillation_loss": 0.06733982264995575,
      "epoch": 1.7,
      "learning_rate": 4.306324316833203e-05,
      "loss": 0.0625,
      "step": 1794,
      "task_loss": 0.019262997433543205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7512487774983488,
      "compression_loss": 0.0,
      "distillation_loss": 0.1413702815771103,
      "epoch": 1.7,
      "learning_rate": 4.3055875131882204e-05,
      "loss": 0.1334,
      "step": 1795,
      "task_loss": 0.061707641929388046
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7512708335480076,
      "compression_loss": 0.0,
      "distillation_loss": 0.1853601038455963,
      "epoch": 1.71,
      "learning_rate": 4.30485038154957e-05,
      "loss": 0.1838,
      "step": 1796,
      "task_loss": 0.16993498802185059
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7512928829442577,
      "compression_loss": 0.0,
      "distillation_loss": 0.15853723883628845,
      "epoch": 1.71,
      "learning_rate": 4.304112922051155e-05,
      "loss": 0.1542,
      "step": 1797,
      "task_loss": 0.11558166146278381
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7513149256881029,
      "compression_loss": 0.0,
      "distillation_loss": 0.1695374995470047,
      "epoch": 1.71,
      "learning_rate": 4.30337513482694e-05,
      "loss": 0.1642,
      "step": 1798,
      "task_loss": 0.11586904525756836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7513369617805469,
      "compression_loss": 0.0,
      "distillation_loss": 0.05897611379623413,
      "epoch": 1.71,
      "learning_rate": 4.3026370200109463e-05,
      "loss": 0.0712,
      "step": 1799,
      "task_loss": 0.18152813613414764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7513589912225933,
      "compression_loss": 0.0,
      "distillation_loss": 0.12656652927398682,
      "epoch": 1.71,
      "learning_rate": 4.301898577737255e-05,
      "loss": 0.1314,
      "step": 1800,
      "task_loss": 0.17490315437316895
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7513810140152459,
      "compression_loss": 0.0,
      "distillation_loss": 0.05648940056562424,
      "epoch": 1.71,
      "learning_rate": 4.3011598081400105e-05,
      "loss": 0.0567,
      "step": 1801,
      "task_loss": 0.05847236514091492
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7514030301595083,
      "compression_loss": 0.0,
      "distillation_loss": 0.20593807101249695,
      "epoch": 1.71,
      "learning_rate": 4.3004207113534124e-05,
      "loss": 0.1972,
      "step": 1802,
      "task_loss": 0.1184128075838089
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7514250396563842,
      "compression_loss": 0.0,
      "distillation_loss": 0.11276452243328094,
      "epoch": 1.71,
      "learning_rate": 4.2996812875117206e-05,
      "loss": 0.1387,
      "step": 1803,
      "task_loss": 0.3719395101070404
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7514470425068772,
      "compression_loss": 0.0,
      "distillation_loss": 0.23946917057037354,
      "epoch": 1.71,
      "learning_rate": 4.2989415367492556e-05,
      "loss": 0.2358,
      "step": 1804,
      "task_loss": 0.20264212787151337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7514690387119911,
      "compression_loss": 0.0,
      "distillation_loss": 0.14112165570259094,
      "epoch": 1.71,
      "learning_rate": 4.298201459200397e-05,
      "loss": 0.1457,
      "step": 1805,
      "task_loss": 0.1867811530828476
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7514910282727296,
      "compression_loss": 0.0,
      "distillation_loss": 0.18797874450683594,
      "epoch": 1.72,
      "learning_rate": 4.2974610549995834e-05,
      "loss": 0.187,
      "step": 1806,
      "task_loss": 0.17826765775680542
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7515130111900963,
      "compression_loss": 0.0,
      "distillation_loss": 0.16024428606033325,
      "epoch": 1.72,
      "learning_rate": 4.296720324281311e-05,
      "loss": 0.157,
      "step": 1807,
      "task_loss": 0.12820284068584442
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.751534987465095,
      "compression_loss": 0.0,
      "distillation_loss": 0.16456547379493713,
      "epoch": 1.72,
      "learning_rate": 4.29597926718014e-05,
      "loss": 0.1509,
      "step": 1808,
      "task_loss": 0.028283601626753807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.751556957098729,
      "compression_loss": 0.0,
      "distillation_loss": 0.085496686398983,
      "epoch": 1.72,
      "learning_rate": 4.295237883830685e-05,
      "loss": 0.0912,
      "step": 1809,
      "task_loss": 0.14275582134723663
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7515789200920026,
      "compression_loss": 0.0,
      "distillation_loss": 0.24256683886051178,
      "epoch": 1.72,
      "learning_rate": 4.294496174367623e-05,
      "loss": 0.2418,
      "step": 1810,
      "task_loss": 0.23468558490276337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.751600876445919,
      "compression_loss": 0.0,
      "distillation_loss": 0.08293971419334412,
      "epoch": 1.72,
      "learning_rate": 4.2937541389256877e-05,
      "loss": 0.0832,
      "step": 1811,
      "task_loss": 0.08516831696033478
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.751622826161482,
      "compression_loss": 0.0,
      "distillation_loss": 0.15826871991157532,
      "epoch": 1.72,
      "learning_rate": 4.293011777639675e-05,
      "loss": 0.1666,
      "step": 1812,
      "task_loss": 0.2412494570016861
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7516447692396955,
      "compression_loss": 0.0,
      "distillation_loss": 0.16091185808181763,
      "epoch": 1.72,
      "learning_rate": 4.2922690906444374e-05,
      "loss": 0.1583,
      "step": 1813,
      "task_loss": 0.13493165373802185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7516667056815628,
      "compression_loss": 0.0,
      "distillation_loss": 0.06874893605709076,
      "epoch": 1.72,
      "learning_rate": 4.291526078074888e-05,
      "loss": 0.0689,
      "step": 1814,
      "task_loss": 0.07061054557561874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7516886354880878,
      "compression_loss": 0.0,
      "distillation_loss": 0.1136234849691391,
      "epoch": 1.72,
      "learning_rate": 4.290782740065997e-05,
      "loss": 0.1169,
      "step": 1815,
      "task_loss": 0.14640654623508453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7517105586602743,
      "compression_loss": 0.0,
      "distillation_loss": 0.14775055646896362,
      "epoch": 1.72,
      "learning_rate": 4.290039076752799e-05,
      "loss": 0.1389,
      "step": 1816,
      "task_loss": 0.059689588844776154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7517324751991258,
      "compression_loss": 0.0,
      "distillation_loss": 0.20547989010810852,
      "epoch": 1.73,
      "learning_rate": 4.28929508827038e-05,
      "loss": 0.2006,
      "step": 1817,
      "task_loss": 0.15708599984645844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.751754385105646,
      "compression_loss": 0.0,
      "distillation_loss": 0.03952891379594803,
      "epoch": 1.73,
      "learning_rate": 4.288550774753892e-05,
      "loss": 0.0444,
      "step": 1818,
      "task_loss": 0.08782260119915009
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7517762883808385,
      "compression_loss": 0.0,
      "distillation_loss": 0.0741434320807457,
      "epoch": 1.73,
      "learning_rate": 4.2878061363385414e-05,
      "loss": 0.0734,
      "step": 1819,
      "task_loss": 0.06647836416959763
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7517981850257073,
      "compression_loss": 0.0,
      "distillation_loss": 0.08467371761798859,
      "epoch": 1.73,
      "learning_rate": 4.287061173159597e-05,
      "loss": 0.0828,
      "step": 1820,
      "task_loss": 0.06570696830749512
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7518200750412558,
      "compression_loss": 0.0,
      "distillation_loss": 0.14830920100212097,
      "epoch": 1.73,
      "learning_rate": 4.286315885352382e-05,
      "loss": 0.1399,
      "step": 1821,
      "task_loss": 0.06412569433450699
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7518419584284877,
      "compression_loss": 0.0,
      "distillation_loss": 0.13605275750160217,
      "epoch": 1.73,
      "learning_rate": 4.285570273052285e-05,
      "loss": 0.1302,
      "step": 1822,
      "task_loss": 0.07789816707372665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7518638351884068,
      "compression_loss": 0.0,
      "distillation_loss": 0.12208747863769531,
      "epoch": 1.73,
      "learning_rate": 4.2848243363947484e-05,
      "loss": 0.1132,
      "step": 1823,
      "task_loss": 0.032726749777793884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7518857053220167,
      "compression_loss": 0.0,
      "distillation_loss": 0.08085359632968903,
      "epoch": 1.73,
      "learning_rate": 4.2840780755152746e-05,
      "loss": 0.0879,
      "step": 1824,
      "task_loss": 0.1517452895641327
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7519075688303211,
      "compression_loss": 0.0,
      "distillation_loss": 0.22030451893806458,
      "epoch": 1.73,
      "learning_rate": 4.283331490549426e-05,
      "loss": 0.2232,
      "step": 1825,
      "task_loss": 0.24884700775146484
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7519294257143236,
      "compression_loss": 0.0,
      "distillation_loss": 0.23822423815727234,
      "epoch": 1.73,
      "learning_rate": 4.282584581632824e-05,
      "loss": 0.2456,
      "step": 1826,
      "task_loss": 0.3118036389350891
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7519512759750281,
      "compression_loss": 0.0,
      "distillation_loss": 0.062073417007923126,
      "epoch": 1.74,
      "learning_rate": 4.281837348901148e-05,
      "loss": 0.0686,
      "step": 1827,
      "task_loss": 0.12697717547416687
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7519731196134382,
      "compression_loss": 0.0,
      "distillation_loss": 0.03941449895501137,
      "epoch": 1.74,
      "learning_rate": 4.281089792490136e-05,
      "loss": 0.0391,
      "step": 1828,
      "task_loss": 0.036043956875801086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7519949566305574,
      "compression_loss": 0.0,
      "distillation_loss": 0.14730878174304962,
      "epoch": 1.74,
      "learning_rate": 4.280341912535585e-05,
      "loss": 0.1414,
      "step": 1829,
      "task_loss": 0.08822986483573914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7520167870273895,
      "compression_loss": 0.0,
      "distillation_loss": 0.2598916292190552,
      "epoch": 1.74,
      "learning_rate": 4.2795937091733515e-05,
      "loss": 0.2527,
      "step": 1830,
      "task_loss": 0.18784040212631226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7520386108049383,
      "compression_loss": 0.0,
      "distillation_loss": 0.0485055074095726,
      "epoch": 1.74,
      "learning_rate": 4.27884518253935e-05,
      "loss": 0.0557,
      "step": 1831,
      "task_loss": 0.12016290426254272
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7520604279642074,
      "compression_loss": 0.0,
      "distillation_loss": 0.3198782205581665,
      "epoch": 1.74,
      "learning_rate": 4.278096332769555e-05,
      "loss": 0.3177,
      "step": 1832,
      "task_loss": 0.29813024401664734
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7520822385062003,
      "compression_loss": 0.0,
      "distillation_loss": 0.10080970823764801,
      "epoch": 1.74,
      "learning_rate": 4.277347159999997e-05,
      "loss": 0.1022,
      "step": 1833,
      "task_loss": 0.11454527825117111
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.752104042431921,
      "compression_loss": 0.0,
      "distillation_loss": 0.06464186310768127,
      "epoch": 1.74,
      "learning_rate": 4.276597664366767e-05,
      "loss": 0.0685,
      "step": 1834,
      "task_loss": 0.1035253256559372
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7521258397423729,
      "compression_loss": 0.0,
      "distillation_loss": 0.2522902190685272,
      "epoch": 1.74,
      "learning_rate": 4.2758478460060166e-05,
      "loss": 0.2398,
      "step": 1835,
      "task_loss": 0.12788613140583038
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7521476304385599,
      "compression_loss": 0.0,
      "distillation_loss": 0.15172149240970612,
      "epoch": 1.74,
      "learning_rate": 4.275097705053951e-05,
      "loss": 0.1433,
      "step": 1836,
      "task_loss": 0.06708873808383942
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7521694145214856,
      "compression_loss": 0.0,
      "distillation_loss": 0.08506453782320023,
      "epoch": 1.74,
      "learning_rate": 4.2743472416468385e-05,
      "loss": 0.0814,
      "step": 1837,
      "task_loss": 0.048387959599494934
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7521911919921536,
      "compression_loss": 0.0,
      "distillation_loss": 0.07385560125112534,
      "epoch": 1.75,
      "learning_rate": 4.2735964559210054e-05,
      "loss": 0.0693,
      "step": 1838,
      "task_loss": 0.028625313192605972
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7522129628515677,
      "compression_loss": 0.0,
      "distillation_loss": 0.16923248767852783,
      "epoch": 1.75,
      "learning_rate": 4.272845348012833e-05,
      "loss": 0.1628,
      "step": 1839,
      "task_loss": 0.10537364333868027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7522347271007315,
      "compression_loss": 0.0,
      "distillation_loss": 0.3296600878238678,
      "epoch": 1.75,
      "learning_rate": 4.272093918058766e-05,
      "loss": 0.3125,
      "step": 1840,
      "task_loss": 0.1578628420829773
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7522564847406487,
      "compression_loss": 0.0,
      "distillation_loss": 0.17389464378356934,
      "epoch": 1.75,
      "learning_rate": 4.271342166195304e-05,
      "loss": 0.173,
      "step": 1841,
      "task_loss": 0.1645306646823883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.752278235772323,
      "compression_loss": 0.0,
      "distillation_loss": 0.08731499314308167,
      "epoch": 1.75,
      "learning_rate": 4.2705900925590056e-05,
      "loss": 0.081,
      "step": 1842,
      "task_loss": 0.024491865187883377
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7522999801967581,
      "compression_loss": 0.0,
      "distillation_loss": 0.13219568133354187,
      "epoch": 1.75,
      "learning_rate": 4.269837697286491e-05,
      "loss": 0.1257,
      "step": 1843,
      "task_loss": 0.06739476323127747
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7523217180149577,
      "compression_loss": 0.0,
      "distillation_loss": 0.06735164672136307,
      "epoch": 1.75,
      "learning_rate": 4.269084980514434e-05,
      "loss": 0.0733,
      "step": 1844,
      "task_loss": 0.12667877972126007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7523434492279254,
      "compression_loss": 0.0,
      "distillation_loss": 0.1707264930009842,
      "epoch": 1.75,
      "learning_rate": 4.268331942379571e-05,
      "loss": 0.1621,
      "step": 1845,
      "task_loss": 0.08454715460538864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.752365173836665,
      "compression_loss": 0.0,
      "distillation_loss": 0.15691004693508148,
      "epoch": 1.75,
      "learning_rate": 4.267578583018694e-05,
      "loss": 0.1498,
      "step": 1846,
      "task_loss": 0.0854022279381752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.75238689184218,
      "compression_loss": 0.0,
      "distillation_loss": 0.17727318406105042,
      "epoch": 1.75,
      "learning_rate": 4.2668249025686545e-05,
      "loss": 0.1772,
      "step": 1847,
      "task_loss": 0.1768154799938202
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7524086032454743,
      "compression_loss": 0.0,
      "distillation_loss": 0.14297039806842804,
      "epoch": 1.75,
      "learning_rate": 4.2660709011663624e-05,
      "loss": 0.1431,
      "step": 1848,
      "task_loss": 0.14434432983398438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7524303080475514,
      "compression_loss": 0.0,
      "distillation_loss": 0.12875759601593018,
      "epoch": 1.76,
      "learning_rate": 4.2653165789487864e-05,
      "loss": 0.128,
      "step": 1849,
      "task_loss": 0.12122198939323425
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7524520062494151,
      "compression_loss": 0.0,
      "distillation_loss": 0.048097141087055206,
      "epoch": 1.76,
      "learning_rate": 4.2645619360529514e-05,
      "loss": 0.046,
      "step": 1850,
      "task_loss": 0.02708207257091999
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.752473697852069,
      "compression_loss": 0.0,
      "distillation_loss": 0.04155917465686798,
      "epoch": 1.76,
      "learning_rate": 4.2638069726159424e-05,
      "loss": 0.0424,
      "step": 1851,
      "task_loss": 0.04985608160495758
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7524953828565168,
      "compression_loss": 0.0,
      "distillation_loss": 0.06510326266288757,
      "epoch": 1.76,
      "learning_rate": 4.263051688774902e-05,
      "loss": 0.0617,
      "step": 1852,
      "task_loss": 0.03152014687657356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7525170612637623,
      "compression_loss": 0.0,
      "distillation_loss": 0.03942890465259552,
      "epoch": 1.76,
      "learning_rate": 4.262296084667032e-05,
      "loss": 0.0426,
      "step": 1853,
      "task_loss": 0.07090801745653152
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.752538733074809,
      "compression_loss": 0.0,
      "distillation_loss": 0.02327614463865757,
      "epoch": 1.76,
      "learning_rate": 4.2615401604295905e-05,
      "loss": 0.0213,
      "step": 1854,
      "task_loss": 0.0035562757402658463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7525603982906607,
      "compression_loss": 0.0,
      "distillation_loss": 0.20073172450065613,
      "epoch": 1.76,
      "learning_rate": 4.260783916199895e-05,
      "loss": 0.1898,
      "step": 1855,
      "task_loss": 0.09144563972949982
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.752582056912321,
      "compression_loss": 0.0,
      "distillation_loss": 0.12031463533639908,
      "epoch": 1.76,
      "learning_rate": 4.260027352115321e-05,
      "loss": 0.1236,
      "step": 1856,
      "task_loss": 0.15320178866386414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7526037089407938,
      "compression_loss": 0.0,
      "distillation_loss": 0.1642073094844818,
      "epoch": 1.76,
      "learning_rate": 4.2592704683133035e-05,
      "loss": 0.1642,
      "step": 1857,
      "task_loss": 0.16391614079475403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7526253543770824,
      "compression_loss": 0.0,
      "distillation_loss": 0.041345108300447464,
      "epoch": 1.76,
      "learning_rate": 4.258513264931331e-05,
      "loss": 0.0553,
      "step": 1858,
      "task_loss": 0.18059666454792023
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7526469932221909,
      "compression_loss": 0.0,
      "distillation_loss": 0.41769933700561523,
      "epoch": 1.77,
      "learning_rate": 4.257755742106956e-05,
      "loss": 0.4052,
      "step": 1859,
      "task_loss": 0.29252398014068604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7526686254771225,
      "compression_loss": 0.0,
      "distillation_loss": 0.32178372144699097,
      "epoch": 1.77,
      "learning_rate": 4.256997899977784e-05,
      "loss": 0.3091,
      "step": 1860,
      "task_loss": 0.19510038197040558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7526902511428814,
      "compression_loss": 0.0,
      "distillation_loss": 0.13911914825439453,
      "epoch": 1.77,
      "learning_rate": 4.2562397386814823e-05,
      "loss": 0.133,
      "step": 1861,
      "task_loss": 0.07751937210559845
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7527118702204709,
      "compression_loss": 0.0,
      "distillation_loss": 0.04612383246421814,
      "epoch": 1.77,
      "learning_rate": 4.255481258355773e-05,
      "loss": 0.0521,
      "step": 1862,
      "task_loss": 0.10636930912733078
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7527334827108949,
      "compression_loss": 0.0,
      "distillation_loss": 0.1954609900712967,
      "epoch": 1.77,
      "learning_rate": 4.254722459138441e-05,
      "loss": 0.1994,
      "step": 1863,
      "task_loss": 0.23512773215770721
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.752755088615157,
      "compression_loss": 0.0,
      "distillation_loss": 0.29851043224334717,
      "epoch": 1.77,
      "learning_rate": 4.253963341167321e-05,
      "loss": 0.2852,
      "step": 1864,
      "task_loss": 0.16501733660697937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7527766879342609,
      "compression_loss": 0.0,
      "distillation_loss": 0.10969699919223785,
      "epoch": 1.77,
      "learning_rate": 4.253203904580314e-05,
      "loss": 0.1038,
      "step": 1865,
      "task_loss": 0.050924863666296005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7527982806692103,
      "compression_loss": 0.0,
      "distillation_loss": 0.1796002984046936,
      "epoch": 1.77,
      "learning_rate": 4.252444149515374e-05,
      "loss": 0.177,
      "step": 1866,
      "task_loss": 0.15360799431800842
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7528198668210088,
      "compression_loss": 0.0,
      "distillation_loss": 0.14553160965442657,
      "epoch": 1.77,
      "learning_rate": 4.251684076110514e-05,
      "loss": 0.1432,
      "step": 1867,
      "task_loss": 0.122085340321064
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7528414463906601,
      "compression_loss": 0.0,
      "distillation_loss": 0.2174375355243683,
      "epoch": 1.77,
      "learning_rate": 4.250923684503806e-05,
      "loss": 0.207,
      "step": 1868,
      "task_loss": 0.11338604241609573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7528630193791679,
      "compression_loss": 0.0,
      "distillation_loss": 0.29796043038368225,
      "epoch": 1.77,
      "learning_rate": 4.2501629748333774e-05,
      "loss": 0.2902,
      "step": 1869,
      "task_loss": 0.22043783962726593
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7528845857875359,
      "compression_loss": 0.0,
      "distillation_loss": 0.12145286798477173,
      "epoch": 1.78,
      "learning_rate": 4.249401947237417e-05,
      "loss": 0.1337,
      "step": 1870,
      "task_loss": 0.24400418996810913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7529061456167679,
      "compression_loss": 0.0,
      "distillation_loss": 0.1129135936498642,
      "epoch": 1.78,
      "learning_rate": 4.248640601854166e-05,
      "loss": 0.1107,
      "step": 1871,
      "task_loss": 0.09064217656850815
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7529276988678674,
      "compression_loss": 0.0,
      "distillation_loss": 0.2056802213191986,
      "epoch": 1.78,
      "learning_rate": 4.247878938821929e-05,
      "loss": 0.2075,
      "step": 1872,
      "task_loss": 0.22419533133506775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7529492455418382,
      "compression_loss": 0.0,
      "distillation_loss": 0.14987275004386902,
      "epoch": 1.78,
      "learning_rate": 4.247116958279065e-05,
      "loss": 0.1442,
      "step": 1873,
      "task_loss": 0.09318174421787262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7529707856396838,
      "compression_loss": 0.0,
      "distillation_loss": 0.09295307099819183,
      "epoch": 1.78,
      "learning_rate": 4.246354660363991e-05,
      "loss": 0.0863,
      "step": 1874,
      "task_loss": 0.026577245444059372
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.752992319162408,
      "compression_loss": 0.0,
      "distillation_loss": 0.10765902698040009,
      "epoch": 1.78,
      "learning_rate": 4.245592045215182e-05,
      "loss": 0.1006,
      "step": 1875,
      "task_loss": 0.03689169883728027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7530138461110146,
      "compression_loss": 0.0,
      "distillation_loss": 0.08064229786396027,
      "epoch": 1.78,
      "learning_rate": 4.244829112971172e-05,
      "loss": 0.0793,
      "step": 1876,
      "task_loss": 0.06695515662431717
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7530353664865072,
      "compression_loss": 0.0,
      "distillation_loss": 0.07568307220935822,
      "epoch": 1.78,
      "learning_rate": 4.24406586377055e-05,
      "loss": 0.0758,
      "step": 1877,
      "task_loss": 0.07722204923629761
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7530568802898894,
      "compression_loss": 0.0,
      "distillation_loss": 0.16826829314231873,
      "epoch": 1.78,
      "learning_rate": 4.2433022977519645e-05,
      "loss": 0.1746,
      "step": 1878,
      "task_loss": 0.23110389709472656
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7530783875221648,
      "compression_loss": 0.0,
      "distillation_loss": 0.22350825369358063,
      "epoch": 1.78,
      "learning_rate": 4.2425384150541206e-05,
      "loss": 0.2159,
      "step": 1879,
      "task_loss": 0.14728033542633057
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7530998881843374,
      "compression_loss": 0.0,
      "distillation_loss": 0.13128520548343658,
      "epoch": 1.79,
      "learning_rate": 4.2417742158157816e-05,
      "loss": 0.1253,
      "step": 1880,
      "task_loss": 0.07127057015895844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7531213822774107,
      "compression_loss": 0.0,
      "distillation_loss": 0.062282584607601166,
      "epoch": 1.79,
      "learning_rate": 4.2410097001757676e-05,
      "loss": 0.0574,
      "step": 1881,
      "task_loss": 0.013064134865999222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7531428698023883,
      "compression_loss": 0.0,
      "distillation_loss": 0.014240816235542297,
      "epoch": 1.79,
      "learning_rate": 4.2402448682729566e-05,
      "loss": 0.0133,
      "step": 1882,
      "task_loss": 0.005121858790516853
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.753164350760274,
      "compression_loss": 0.0,
      "distillation_loss": 0.08013554662466049,
      "epoch": 1.79,
      "learning_rate": 4.2394797202462844e-05,
      "loss": 0.074,
      "step": 1883,
      "task_loss": 0.019022256135940552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7531858251520714,
      "compression_loss": 0.0,
      "distillation_loss": 0.3540295362472534,
      "epoch": 1.79,
      "learning_rate": 4.238714256234744e-05,
      "loss": 0.3391,
      "step": 1884,
      "task_loss": 0.20519450306892395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7532072929787843,
      "compression_loss": 0.0,
      "distillation_loss": 0.21610870957374573,
      "epoch": 1.79,
      "learning_rate": 4.237948476377385e-05,
      "loss": 0.2104,
      "step": 1885,
      "task_loss": 0.15874995291233063
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7532287542414162,
      "compression_loss": 0.0,
      "distillation_loss": 0.12253059446811676,
      "epoch": 1.79,
      "learning_rate": 4.237182380813315e-05,
      "loss": 0.1246,
      "step": 1886,
      "task_loss": 0.14349707961082458
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.753250208940971,
      "compression_loss": 0.0,
      "distillation_loss": 0.02185000665485859,
      "epoch": 1.79,
      "learning_rate": 4.236415969681699e-05,
      "loss": 0.0202,
      "step": 1887,
      "task_loss": 0.00525074265897274
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7532716570784523,
      "compression_loss": 0.0,
      "distillation_loss": 0.1840369701385498,
      "epoch": 1.79,
      "learning_rate": 4.23564924312176e-05,
      "loss": 0.1743,
      "step": 1888,
      "task_loss": 0.08696576207876205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7532930986548637,
      "compression_loss": 0.0,
      "distillation_loss": 0.23074445128440857,
      "epoch": 1.79,
      "learning_rate": 4.2348822012727765e-05,
      "loss": 0.2274,
      "step": 1889,
      "task_loss": 0.19752009212970734
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7533145336712089,
      "compression_loss": 0.0,
      "distillation_loss": 0.04407363384962082,
      "epoch": 1.79,
      "learning_rate": 4.234114844274086e-05,
      "loss": 0.0408,
      "step": 1890,
      "task_loss": 0.01094069704413414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7533359621284917,
      "compression_loss": 0.0,
      "distillation_loss": 0.11152929067611694,
      "epoch": 1.8,
      "learning_rate": 4.2333471722650826e-05,
      "loss": 0.1117,
      "step": 1891,
      "task_loss": 0.11322666704654694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7533573840277157,
      "compression_loss": 0.0,
      "distillation_loss": 0.09612832963466644,
      "epoch": 1.8,
      "learning_rate": 4.232579185385217e-05,
      "loss": 0.0905,
      "step": 1892,
      "task_loss": 0.03963814303278923
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7533787993698846,
      "compression_loss": 0.0,
      "distillation_loss": 0.3889869451522827,
      "epoch": 1.8,
      "learning_rate": 4.231810883773999e-05,
      "loss": 0.3714,
      "step": 1893,
      "task_loss": 0.2126784771680832
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.753400208156002,
      "compression_loss": 0.0,
      "distillation_loss": 0.02580692060291767,
      "epoch": 1.8,
      "learning_rate": 4.231042267570993e-05,
      "loss": 0.0241,
      "step": 1894,
      "task_loss": 0.008529577404260635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7534216103870717,
      "compression_loss": 0.0,
      "distillation_loss": 0.03107302263379097,
      "epoch": 1.8,
      "learning_rate": 4.230273336915822e-05,
      "loss": 0.0289,
      "step": 1895,
      "task_loss": 0.0094615388661623
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7534430060640973,
      "compression_loss": 0.0,
      "distillation_loss": 0.03097601607441902,
      "epoch": 1.8,
      "learning_rate": 4.2295040919481664e-05,
      "loss": 0.0287,
      "step": 1896,
      "task_loss": 0.007949141785502434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7534643951880825,
      "compression_loss": 0.0,
      "distillation_loss": 0.05745317414402962,
      "epoch": 1.8,
      "learning_rate": 4.228734532807763e-05,
      "loss": 0.0657,
      "step": 1897,
      "task_loss": 0.13991054892539978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.753485777760031,
      "compression_loss": 0.0,
      "distillation_loss": 0.028573032468557358,
      "epoch": 1.8,
      "learning_rate": 4.2279646596344067e-05,
      "loss": 0.0357,
      "step": 1898,
      "task_loss": 0.09934914112091064
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7535071537809466,
      "compression_loss": 0.0,
      "distillation_loss": 0.18852916359901428,
      "epoch": 1.8,
      "learning_rate": 4.227194472567948e-05,
      "loss": 0.1943,
      "step": 1899,
      "task_loss": 0.24627923965454102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7535285232518327,
      "compression_loss": 0.0,
      "distillation_loss": 0.24925187230110168,
      "epoch": 1.8,
      "learning_rate": 4.2264239717482945e-05,
      "loss": 0.2396,
      "step": 1900,
      "task_loss": 0.15310557186603546
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7535498861736932,
      "compression_loss": 0.0,
      "distillation_loss": 0.0920097678899765,
      "epoch": 1.81,
      "learning_rate": 4.225653157315412e-05,
      "loss": 0.0899,
      "step": 1901,
      "task_loss": 0.07052158564329147
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7535712425475317,
      "compression_loss": 0.0,
      "distillation_loss": 0.13238155841827393,
      "epoch": 1.81,
      "learning_rate": 4.224882029409323e-05,
      "loss": 0.1305,
      "step": 1902,
      "task_loss": 0.1139896884560585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7535925923743519,
      "compression_loss": 0.0,
      "distillation_loss": 0.044050946831703186,
      "epoch": 1.81,
      "learning_rate": 4.224110588170106e-05,
      "loss": 0.0424,
      "step": 1903,
      "task_loss": 0.027057217434048653
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7536139356551576,
      "compression_loss": 0.0,
      "distillation_loss": 0.1628579944372177,
      "epoch": 1.81,
      "learning_rate": 4.223338833737898e-05,
      "loss": 0.154,
      "step": 1904,
      "task_loss": 0.07387073338031769
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7536352723909523,
      "compression_loss": 0.0,
      "distillation_loss": 0.1415494978427887,
      "epoch": 1.81,
      "learning_rate": 4.22256676625289e-05,
      "loss": 0.1376,
      "step": 1905,
      "task_loss": 0.10222296416759491
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7536566025827397,
      "compression_loss": 0.0,
      "distillation_loss": 0.016083184629678726,
      "epoch": 1.81,
      "learning_rate": 4.221794385855334e-05,
      "loss": 0.0149,
      "step": 1906,
      "task_loss": 0.003827514126896858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7536779262315236,
      "compression_loss": 0.0,
      "distillation_loss": 0.02969953790307045,
      "epoch": 1.81,
      "learning_rate": 4.221021692685534e-05,
      "loss": 0.0271,
      "step": 1907,
      "task_loss": 0.0035870037972927094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7536992433383076,
      "compression_loss": 0.0,
      "distillation_loss": 0.13226908445358276,
      "epoch": 1.81,
      "learning_rate": 4.220248686883857e-05,
      "loss": 0.1333,
      "step": 1908,
      "task_loss": 0.14230315387248993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7537205539040954,
      "compression_loss": 0.0,
      "distillation_loss": 0.09493161737918854,
      "epoch": 1.81,
      "learning_rate": 4.21947536859072e-05,
      "loss": 0.0904,
      "step": 1909,
      "task_loss": 0.04967789724469185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7537418579298907,
      "compression_loss": 0.0,
      "distillation_loss": 0.2932942509651184,
      "epoch": 1.81,
      "learning_rate": 4.218701737946601e-05,
      "loss": 0.2855,
      "step": 1910,
      "task_loss": 0.21534579992294312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7537631554166971,
      "compression_loss": 0.0,
      "distillation_loss": 0.2947857081890106,
      "epoch": 1.81,
      "learning_rate": 4.217927795092034e-05,
      "loss": 0.2831,
      "step": 1911,
      "task_loss": 0.17762699723243713
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7537844463655184,
      "compression_loss": 0.0,
      "distillation_loss": 0.2603399455547333,
      "epoch": 1.82,
      "learning_rate": 4.21715354016761e-05,
      "loss": 0.2464,
      "step": 1912,
      "task_loss": 0.12051122635602951
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7538057307773582,
      "compression_loss": 0.0,
      "distillation_loss": 0.19969797134399414,
      "epoch": 1.82,
      "learning_rate": 4.216378973313976e-05,
      "loss": 0.1898,
      "step": 1913,
      "task_loss": 0.10046619176864624
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7538270086532203,
      "compression_loss": 0.0,
      "distillation_loss": 0.08806608617305756,
      "epoch": 1.82,
      "learning_rate": 4.215604094671835e-05,
      "loss": 0.0839,
      "step": 1914,
      "task_loss": 0.04651288688182831
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7538482799941082,
      "compression_loss": 0.0,
      "distillation_loss": 0.09663443267345428,
      "epoch": 1.82,
      "learning_rate": 4.214828904381947e-05,
      "loss": 0.1037,
      "step": 1915,
      "task_loss": 0.16719813644886017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7538695448010257,
      "compression_loss": 0.0,
      "distillation_loss": 0.20862799882888794,
      "epoch": 1.82,
      "learning_rate": 4.21405340258513e-05,
      "loss": 0.2019,
      "step": 1916,
      "task_loss": 0.14169104397296906
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7538908030749765,
      "compression_loss": 0.0,
      "distillation_loss": 0.11301036179065704,
      "epoch": 1.82,
      "learning_rate": 4.213277589422258e-05,
      "loss": 0.117,
      "step": 1917,
      "task_loss": 0.1530938744544983
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7539120548169642,
      "compression_loss": 0.0,
      "distillation_loss": 0.3912833333015442,
      "epoch": 1.82,
      "learning_rate": 4.21250146503426e-05,
      "loss": 0.388,
      "step": 1918,
      "task_loss": 0.358115017414093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7539333000279925,
      "compression_loss": 0.0,
      "distillation_loss": 0.057676736265420914,
      "epoch": 1.82,
      "learning_rate": 4.2117250295621235e-05,
      "loss": 0.0556,
      "step": 1919,
      "task_loss": 0.03697335347533226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7539545387090651,
      "compression_loss": 0.0,
      "distillation_loss": 0.08969536423683167,
      "epoch": 1.82,
      "learning_rate": 4.210948283146892e-05,
      "loss": 0.0894,
      "step": 1920,
      "task_loss": 0.08631106466054916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7539757708611857,
      "compression_loss": 0.0,
      "distillation_loss": 0.17513608932495117,
      "epoch": 1.82,
      "learning_rate": 4.210171225929664e-05,
      "loss": 0.1674,
      "step": 1921,
      "task_loss": 0.09771417081356049
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.753996996485358,
      "compression_loss": 0.0,
      "distillation_loss": 0.1546175479888916,
      "epoch": 1.83,
      "learning_rate": 4.209393858051598e-05,
      "loss": 0.1516,
      "step": 1922,
      "task_loss": 0.12429603934288025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7540182155825856,
      "compression_loss": 0.0,
      "distillation_loss": 0.28697076439857483,
      "epoch": 1.83,
      "learning_rate": 4.208616179653903e-05,
      "loss": 0.2767,
      "step": 1923,
      "task_loss": 0.18397051095962524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7540394281538723,
      "compression_loss": 0.0,
      "distillation_loss": 0.04862765967845917,
      "epoch": 1.83,
      "learning_rate": 4.207838190877852e-05,
      "loss": 0.0495,
      "step": 1924,
      "task_loss": 0.05719340965151787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7540606342002216,
      "compression_loss": 0.0,
      "distillation_loss": 0.2369440346956253,
      "epoch": 1.83,
      "learning_rate": 4.2070598918647683e-05,
      "loss": 0.2375,
      "step": 1925,
      "task_loss": 0.2421676218509674
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7540818337226374,
      "compression_loss": 0.0,
      "distillation_loss": 0.10661203414201736,
      "epoch": 1.83,
      "learning_rate": 4.206281282756034e-05,
      "loss": 0.101,
      "step": 1926,
      "task_loss": 0.050882913172245026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7541030267221233,
      "compression_loss": 0.0,
      "distillation_loss": 0.09290862083435059,
      "epoch": 1.83,
      "learning_rate": 4.205502363693087e-05,
      "loss": 0.0886,
      "step": 1927,
      "task_loss": 0.04993749409914017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7541242131996828,
      "compression_loss": 0.0,
      "distillation_loss": 0.13861539959907532,
      "epoch": 1.83,
      "learning_rate": 4.204723134817422e-05,
      "loss": 0.1414,
      "step": 1928,
      "task_loss": 0.16668032109737396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7541453931563199,
      "compression_loss": 0.0,
      "distillation_loss": 0.1590609848499298,
      "epoch": 1.83,
      "learning_rate": 4.2039435962705886e-05,
      "loss": 0.1669,
      "step": 1929,
      "task_loss": 0.23708070814609528
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.754166566593038,
      "compression_loss": 0.0,
      "distillation_loss": 0.035407066345214844,
      "epoch": 1.83,
      "learning_rate": 4.2031637481941954e-05,
      "loss": 0.0332,
      "step": 1930,
      "task_loss": 0.013283960521221161
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.754187733510841,
      "compression_loss": 0.0,
      "distillation_loss": 0.1811191439628601,
      "epoch": 1.83,
      "learning_rate": 4.202383590729905e-05,
      "loss": 0.1727,
      "step": 1931,
      "task_loss": 0.09702938795089722
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7542088939107325,
      "compression_loss": 0.0,
      "distillation_loss": 0.10961516201496124,
      "epoch": 1.83,
      "learning_rate": 4.201603124019436e-05,
      "loss": 0.1087,
      "step": 1932,
      "task_loss": 0.10052899271249771
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.754230047793716,
      "compression_loss": 0.0,
      "distillation_loss": 0.249653160572052,
      "epoch": 1.84,
      "learning_rate": 4.200822348204565e-05,
      "loss": 0.2377,
      "step": 1933,
      "task_loss": 0.12971001863479614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7542511951607956,
      "compression_loss": 0.0,
      "distillation_loss": 0.12614881992340088,
      "epoch": 1.84,
      "learning_rate": 4.200041263427123e-05,
      "loss": 0.1307,
      "step": 1934,
      "task_loss": 0.17151018977165222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7542723360129746,
      "compression_loss": 0.0,
      "distillation_loss": 0.18562328815460205,
      "epoch": 1.84,
      "learning_rate": 4.199259869828998e-05,
      "loss": 0.1845,
      "step": 1935,
      "task_loss": 0.1747492104768753
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7542934703512568,
      "compression_loss": 0.0,
      "distillation_loss": 0.07814265787601471,
      "epoch": 1.84,
      "learning_rate": 4.1984781675521345e-05,
      "loss": 0.0725,
      "step": 1936,
      "task_loss": 0.021998286247253418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7543145981766459,
      "compression_loss": 0.0,
      "distillation_loss": 0.06505021452903748,
      "epoch": 1.84,
      "learning_rate": 4.1976961567385306e-05,
      "loss": 0.0609,
      "step": 1937,
      "task_loss": 0.02330428548157215
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7543357194901457,
      "compression_loss": 0.0,
      "distillation_loss": 0.043614309281110764,
      "epoch": 1.84,
      "learning_rate": 4.1969138375302445e-05,
      "loss": 0.0402,
      "step": 1938,
      "task_loss": 0.009535277262330055
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7543568342927597,
      "compression_loss": 0.0,
      "distillation_loss": 0.12613525986671448,
      "epoch": 1.84,
      "learning_rate": 4.1961312100693874e-05,
      "loss": 0.1151,
      "step": 1939,
      "task_loss": 0.016281738877296448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7543779425854915,
      "compression_loss": 0.0,
      "distillation_loss": 0.23481710255146027,
      "epoch": 1.84,
      "learning_rate": 4.1953482744981274e-05,
      "loss": 0.2348,
      "step": 1940,
      "task_loss": 0.23426470160484314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7543990443693451,
      "compression_loss": 0.0,
      "distillation_loss": 0.17542167007923126,
      "epoch": 1.84,
      "learning_rate": 4.194565030958688e-05,
      "loss": 0.1797,
      "step": 1941,
      "task_loss": 0.2181302309036255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.754420139645324,
      "compression_loss": 0.0,
      "distillation_loss": 0.04640250653028488,
      "epoch": 1.84,
      "learning_rate": 4.19378147959335e-05,
      "loss": 0.0431,
      "step": 1942,
      "task_loss": 0.013633305206894875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7544412284144317,
      "compression_loss": 0.0,
      "distillation_loss": 0.09803412854671478,
      "epoch": 1.85,
      "learning_rate": 4.192997620544449e-05,
      "loss": 0.099,
      "step": 1943,
      "task_loss": 0.10792693495750427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7544623106776723,
      "compression_loss": 0.0,
      "distillation_loss": 0.1254318654537201,
      "epoch": 1.85,
      "learning_rate": 4.192213453954377e-05,
      "loss": 0.1264,
      "step": 1944,
      "task_loss": 0.13558337092399597
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7544833864360492,
      "compression_loss": 0.0,
      "distillation_loss": 0.028305238112807274,
      "epoch": 1.85,
      "learning_rate": 4.19142897996558e-05,
      "loss": 0.026,
      "step": 1945,
      "task_loss": 0.005378095433115959
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7545044556905661,
      "compression_loss": 0.0,
      "distillation_loss": 0.08470214158296585,
      "epoch": 1.85,
      "learning_rate": 4.190644198720563e-05,
      "loss": 0.0819,
      "step": 1946,
      "task_loss": 0.05680760368704796
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7545255184422267,
      "compression_loss": 0.0,
      "distillation_loss": 0.2368120402097702,
      "epoch": 1.85,
      "learning_rate": 4.189859110361886e-05,
      "loss": 0.231,
      "step": 1947,
      "task_loss": 0.17879217863082886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7545465746920347,
      "compression_loss": 0.0,
      "distillation_loss": 0.12444418668746948,
      "epoch": 1.85,
      "learning_rate": 4.189073715032163e-05,
      "loss": 0.1211,
      "step": 1948,
      "task_loss": 0.0914890468120575
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7545676244409938,
      "compression_loss": 0.0,
      "distillation_loss": 0.1032329723238945,
      "epoch": 1.85,
      "learning_rate": 4.188288012874065e-05,
      "loss": 0.1,
      "step": 1949,
      "task_loss": 0.07098895311355591
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7545886676901077,
      "compression_loss": 0.0,
      "distillation_loss": 0.6745878458023071,
      "epoch": 1.85,
      "learning_rate": 4.187502004030318e-05,
      "loss": 0.6498,
      "step": 1950,
      "task_loss": 0.4266480803489685
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.75460970444038,
      "compression_loss": 0.0,
      "distillation_loss": 0.2606114149093628,
      "epoch": 1.85,
      "learning_rate": 4.186715688643705e-05,
      "loss": 0.2448,
      "step": 1951,
      "task_loss": 0.10222074389457703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7546307346928145,
      "compression_loss": 0.0,
      "distillation_loss": 0.07269393652677536,
      "epoch": 1.85,
      "learning_rate": 4.185929066857064e-05,
      "loss": 0.0787,
      "step": 1952,
      "task_loss": 0.1328985095024109
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7546517584484147,
      "compression_loss": 0.0,
      "distillation_loss": 0.03857032582163811,
      "epoch": 1.85,
      "learning_rate": 4.1851421388132886e-05,
      "loss": 0.0352,
      "step": 1953,
      "task_loss": 0.004427826032042503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7546727757081845,
      "compression_loss": 0.0,
      "distillation_loss": 0.025319568812847137,
      "epoch": 1.86,
      "learning_rate": 4.1843549046553284e-05,
      "loss": 0.0232,
      "step": 1954,
      "task_loss": 0.004016607999801636
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7546937864731275,
      "compression_loss": 0.0,
      "distillation_loss": 0.14767573773860931,
      "epoch": 1.86,
      "learning_rate": 4.183567364526186e-05,
      "loss": 0.1589,
      "step": 1955,
      "task_loss": 0.2602040767669678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7547147907442472,
      "compression_loss": 0.0,
      "distillation_loss": 0.19196276366710663,
      "epoch": 1.86,
      "learning_rate": 4.182779518568926e-05,
      "loss": 0.1933,
      "step": 1956,
      "task_loss": 0.2052653729915619
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7547357885225475,
      "compression_loss": 0.0,
      "distillation_loss": 0.168940469622612,
      "epoch": 1.86,
      "learning_rate": 4.181991366926661e-05,
      "loss": 0.1697,
      "step": 1957,
      "task_loss": 0.1767549216747284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7547567798090321,
      "compression_loss": 0.0,
      "distillation_loss": 0.06519413739442825,
      "epoch": 1.86,
      "learning_rate": 4.181202909742564e-05,
      "loss": 0.0633,
      "step": 1958,
      "task_loss": 0.04639093205332756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7547777646047046,
      "compression_loss": 0.0,
      "distillation_loss": 0.23005223274230957,
      "epoch": 1.86,
      "learning_rate": 4.1804141471598604e-05,
      "loss": 0.22,
      "step": 1959,
      "task_loss": 0.1295807808637619
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7547987429105686,
      "compression_loss": 0.0,
      "distillation_loss": 0.15140117704868317,
      "epoch": 1.86,
      "learning_rate": 4.179625079321836e-05,
      "loss": 0.1443,
      "step": 1960,
      "task_loss": 0.08072268962860107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.754819714727628,
      "compression_loss": 0.0,
      "distillation_loss": 0.15924590826034546,
      "epoch": 1.86,
      "learning_rate": 4.1788357063718254e-05,
      "loss": 0.1564,
      "step": 1961,
      "task_loss": 0.1309977024793625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7548406800568862,
      "compression_loss": 0.0,
      "distillation_loss": 0.09538164734840393,
      "epoch": 1.86,
      "learning_rate": 4.178046028453224e-05,
      "loss": 0.091,
      "step": 1962,
      "task_loss": 0.05167919024825096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7548616388993471,
      "compression_loss": 0.0,
      "distillation_loss": 0.12580883502960205,
      "epoch": 1.86,
      "learning_rate": 4.1772560457094795e-05,
      "loss": 0.123,
      "step": 1963,
      "task_loss": 0.09781771153211594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7548825912560143,
      "compression_loss": 0.0,
      "distillation_loss": 0.06123030185699463,
      "epoch": 1.87,
      "learning_rate": 4.1764657582840965e-05,
      "loss": 0.0621,
      "step": 1964,
      "task_loss": 0.069994255900383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7549035371278915,
      "compression_loss": 0.0,
      "distillation_loss": 0.09012752026319504,
      "epoch": 1.87,
      "learning_rate": 4.175675166320635e-05,
      "loss": 0.0856,
      "step": 1965,
      "task_loss": 0.045101843774318695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7549244765159824,
      "compression_loss": 0.0,
      "distillation_loss": 0.18190321326255798,
      "epoch": 1.87,
      "learning_rate": 4.1748842699627094e-05,
      "loss": 0.178,
      "step": 1966,
      "task_loss": 0.14312417805194855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7549454094212906,
      "compression_loss": 0.0,
      "distillation_loss": 0.2552461326122284,
      "epoch": 1.87,
      "learning_rate": 4.17409306935399e-05,
      "loss": 0.2553,
      "step": 1967,
      "task_loss": 0.25603681802749634
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.75496633584482,
      "compression_loss": 0.0,
      "distillation_loss": 0.05597781017422676,
      "epoch": 1.87,
      "learning_rate": 4.173301564638201e-05,
      "loss": 0.0518,
      "step": 1968,
      "task_loss": 0.014608925208449364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.754987255787574,
      "compression_loss": 0.0,
      "distillation_loss": 0.24270398914813995,
      "epoch": 1.87,
      "learning_rate": 4.1725097559591256e-05,
      "loss": 0.2384,
      "step": 1969,
      "task_loss": 0.19955337047576904
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7550081692505564,
      "compression_loss": 0.0,
      "distillation_loss": 0.07123062014579773,
      "epoch": 1.87,
      "learning_rate": 4.1717176434605967e-05,
      "loss": 0.0911,
      "step": 1970,
      "task_loss": 0.2696712017059326
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7550290762347709,
      "compression_loss": 0.0,
      "distillation_loss": 0.07343777269124985,
      "epoch": 1.87,
      "learning_rate": 4.170925227286508e-05,
      "loss": 0.0777,
      "step": 1971,
      "task_loss": 0.11573462188243866
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7550499767412212,
      "compression_loss": 0.0,
      "distillation_loss": 0.45374932885169983,
      "epoch": 1.87,
      "learning_rate": 4.170132507580803e-05,
      "loss": 0.4348,
      "step": 1972,
      "task_loss": 0.26441484689712524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.755070870770911,
      "compression_loss": 0.0,
      "distillation_loss": 0.07927072793245316,
      "epoch": 1.87,
      "learning_rate": 4.1693394844874856e-05,
      "loss": 0.0938,
      "step": 1973,
      "task_loss": 0.22447620332241058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7550917583248438,
      "compression_loss": 0.0,
      "distillation_loss": 0.10215029865503311,
      "epoch": 1.87,
      "learning_rate": 4.1685461581506115e-05,
      "loss": 0.1084,
      "step": 1974,
      "task_loss": 0.16435641050338745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7551126394040235,
      "compression_loss": 0.0,
      "distillation_loss": 0.1036095917224884,
      "epoch": 1.88,
      "learning_rate": 4.167752528714291e-05,
      "loss": 0.096,
      "step": 1975,
      "task_loss": 0.027509452775120735
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7551335140094537,
      "compression_loss": 0.0,
      "distillation_loss": 0.19809332489967346,
      "epoch": 1.88,
      "learning_rate": 4.166958596322692e-05,
      "loss": 0.1908,
      "step": 1976,
      "task_loss": 0.1248994842171669
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.755154382142138,
      "compression_loss": 0.0,
      "distillation_loss": 0.10628242045640945,
      "epoch": 1.88,
      "learning_rate": 4.1661643611200366e-05,
      "loss": 0.1052,
      "step": 1977,
      "task_loss": 0.09579187631607056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7551752438030803,
      "compression_loss": 0.0,
      "distillation_loss": 0.3265897333621979,
      "epoch": 1.88,
      "learning_rate": 4.1653698232506e-05,
      "loss": 0.311,
      "step": 1978,
      "task_loss": 0.17026562988758087
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7551960989932841,
      "compression_loss": 0.0,
      "distillation_loss": 0.44924837350845337,
      "epoch": 1.88,
      "learning_rate": 4.1645749828587145e-05,
      "loss": 0.4407,
      "step": 1979,
      "task_loss": 0.36339446902275085
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7552169477137531,
      "compression_loss": 0.0,
      "distillation_loss": 0.2017005980014801,
      "epoch": 1.88,
      "learning_rate": 4.1637798400887674e-05,
      "loss": 0.1914,
      "step": 1980,
      "task_loss": 0.09869912266731262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7552377899654911,
      "compression_loss": 0.0,
      "distillation_loss": 0.24423138797283173,
      "epoch": 1.88,
      "learning_rate": 4.162984395085198e-05,
      "loss": 0.2326,
      "step": 1981,
      "task_loss": 0.12757530808448792
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7552586257495015,
      "compression_loss": 0.0,
      "distillation_loss": 0.06164749711751938,
      "epoch": 1.88,
      "learning_rate": 4.162188647992506e-05,
      "loss": 0.0573,
      "step": 1982,
      "task_loss": 0.018155789002776146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7552794550667884,
      "compression_loss": 0.0,
      "distillation_loss": 0.3318878710269928,
      "epoch": 1.88,
      "learning_rate": 4.161392598955239e-05,
      "loss": 0.3315,
      "step": 1983,
      "task_loss": 0.328228622674942
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7553002779183551,
      "compression_loss": 0.0,
      "distillation_loss": 0.13821963965892792,
      "epoch": 1.88,
      "learning_rate": 4.160596248118007e-05,
      "loss": 0.1504,
      "step": 1984,
      "task_loss": 0.25973179936408997
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7553210943052054,
      "compression_loss": 0.0,
      "distillation_loss": 0.11968272924423218,
      "epoch": 1.89,
      "learning_rate": 4.159799595625468e-05,
      "loss": 0.1284,
      "step": 1985,
      "task_loss": 0.20697957277297974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7553419042283431,
      "compression_loss": 0.0,
      "distillation_loss": 0.14230716228485107,
      "epoch": 1.89,
      "learning_rate": 4.159002641622338e-05,
      "loss": 0.1497,
      "step": 1986,
      "task_loss": 0.21655035018920898
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7553627076887719,
      "compression_loss": 0.0,
      "distillation_loss": 0.053881511092185974,
      "epoch": 1.89,
      "learning_rate": 4.1582053862533895e-05,
      "loss": 0.0549,
      "step": 1987,
      "task_loss": 0.06384479999542236
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7553835046874953,
      "compression_loss": 0.0,
      "distillation_loss": 0.16238471865653992,
      "epoch": 1.89,
      "learning_rate": 4.157407829663446e-05,
      "loss": 0.1679,
      "step": 1988,
      "task_loss": 0.21752187609672546
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7554042952255171,
      "compression_loss": 0.0,
      "distillation_loss": 0.08790767192840576,
      "epoch": 1.89,
      "learning_rate": 4.1566099719973884e-05,
      "loss": 0.084,
      "step": 1989,
      "task_loss": 0.04870061203837395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7554250793038408,
      "compression_loss": 0.0,
      "distillation_loss": 0.12433084100484848,
      "epoch": 1.89,
      "learning_rate": 4.1558118134001514e-05,
      "loss": 0.1188,
      "step": 1990,
      "task_loss": 0.06881560385227203
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7554458569234704,
      "compression_loss": 0.0,
      "distillation_loss": 0.19472339749336243,
      "epoch": 1.89,
      "learning_rate": 4.155013354016723e-05,
      "loss": 0.1847,
      "step": 1991,
      "task_loss": 0.0943620428442955
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7554666280854094,
      "compression_loss": 0.0,
      "distillation_loss": 0.12036952376365662,
      "epoch": 1.89,
      "learning_rate": 4.154214593992149e-05,
      "loss": 0.1135,
      "step": 1992,
      "task_loss": 0.05154300481081009
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7554873927906615,
      "compression_loss": 0.0,
      "distillation_loss": 0.16188105940818787,
      "epoch": 1.89,
      "learning_rate": 4.1534155334715264e-05,
      "loss": 0.1657,
      "step": 1993,
      "task_loss": 0.20030921697616577
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7555081510402303,
      "compression_loss": 0.0,
      "distillation_loss": 0.03598017245531082,
      "epoch": 1.89,
      "learning_rate": 4.15261617260001e-05,
      "loss": 0.0422,
      "step": 1994,
      "task_loss": 0.09828056395053864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7555289028351195,
      "compression_loss": 0.0,
      "distillation_loss": 0.07038508355617523,
      "epoch": 1.89,
      "learning_rate": 4.151816511522807e-05,
      "loss": 0.0708,
      "step": 1995,
      "task_loss": 0.07408405095338821
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.755549648176333,
      "compression_loss": 0.0,
      "distillation_loss": 0.0785357877612114,
      "epoch": 1.9,
      "learning_rate": 4.151016550385179e-05,
      "loss": 0.0895,
      "step": 1996,
      "task_loss": 0.18811197578907013
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7555703870648742,
      "compression_loss": 0.0,
      "distillation_loss": 0.06988924741744995,
      "epoch": 1.9,
      "learning_rate": 4.150216289332443e-05,
      "loss": 0.0647,
      "step": 1997,
      "task_loss": 0.018136270344257355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7555911195017471,
      "compression_loss": 0.0,
      "distillation_loss": 0.11623725295066833,
      "epoch": 1.9,
      "learning_rate": 4.149415728509971e-05,
      "loss": 0.1145,
      "step": 1998,
      "task_loss": 0.09836913645267487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.755611845487955,
      "compression_loss": 0.0,
      "distillation_loss": 0.0843459814786911,
      "epoch": 1.9,
      "learning_rate": 4.1486148680631875e-05,
      "loss": 0.0788,
      "step": 1999,
      "task_loss": 0.028761208057403564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7556325650245018,
      "compression_loss": 0.0,
      "distillation_loss": 0.16870735585689545,
      "epoch": 1.9,
      "learning_rate": 4.147813708137574e-05,
      "loss": 0.1591,
      "step": 2000,
      "task_loss": 0.07283841073513031
    },
    {
      "epoch": 1.9,
      "eval_accuracy": 0.9013761467889908,
      "eval_loss": 0.3940832018852234,
      "eval_runtime": 18.3803,
      "eval_samples_per_second": 47.442,
      "eval_steps_per_second": 5.93,
      "step": 2000
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7556532781123911,
      "compression_loss": 0.0,
      "distillation_loss": 0.2963669002056122,
      "epoch": 1.9,
      "learning_rate": 4.1470122488786645e-05,
      "loss": 0.2864,
      "step": 2001,
      "task_loss": 0.1968749761581421
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7556739847526267,
      "compression_loss": 0.0,
      "distillation_loss": 0.05983327701687813,
      "epoch": 1.9,
      "learning_rate": 4.146210490432048e-05,
      "loss": 0.0605,
      "step": 2002,
      "task_loss": 0.06697780638933182
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7556946849462123,
      "compression_loss": 0.0,
      "distillation_loss": 0.07671034336090088,
      "epoch": 1.9,
      "learning_rate": 4.1454084329433674e-05,
      "loss": 0.0927,
      "step": 2003,
      "task_loss": 0.23632298409938812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7557153786941514,
      "compression_loss": 0.0,
      "distillation_loss": 0.14522545039653778,
      "epoch": 1.9,
      "learning_rate": 4.144606076558321e-05,
      "loss": 0.1386,
      "step": 2004,
      "task_loss": 0.07935189455747604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7557360659974478,
      "compression_loss": 0.0,
      "distillation_loss": 0.25535622239112854,
      "epoch": 1.9,
      "learning_rate": 4.14380342142266e-05,
      "loss": 0.2454,
      "step": 2005,
      "task_loss": 0.1559627205133438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7557567468571051,
      "compression_loss": 0.0,
      "distillation_loss": 0.09220820665359497,
      "epoch": 1.91,
      "learning_rate": 4.14300046768219e-05,
      "loss": 0.0889,
      "step": 2006,
      "task_loss": 0.05925717204809189
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7557774212741272,
      "compression_loss": 0.0,
      "distillation_loss": 0.21450912952423096,
      "epoch": 1.91,
      "learning_rate": 4.1421972154827724e-05,
      "loss": 0.2048,
      "step": 2007,
      "task_loss": 0.11772306263446808
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7557980892495175,
      "compression_loss": 0.0,
      "distillation_loss": 0.16318204998970032,
      "epoch": 1.91,
      "learning_rate": 4.141393664970323e-05,
      "loss": 0.1539,
      "step": 2008,
      "task_loss": 0.0701281875371933
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7558187507842798,
      "compression_loss": 0.0,
      "distillation_loss": 0.0389268696308136,
      "epoch": 1.91,
      "learning_rate": 4.140589816290808e-05,
      "loss": 0.0364,
      "step": 2009,
      "task_loss": 0.013847017660737038
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7558394058794179,
      "compression_loss": 0.0,
      "distillation_loss": 0.047759756445884705,
      "epoch": 1.91,
      "learning_rate": 4.1397856695902535e-05,
      "loss": 0.085,
      "step": 2010,
      "task_loss": 0.4196968376636505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7558600545359353,
      "compression_loss": 0.0,
      "distillation_loss": 0.0479925200343132,
      "epoch": 1.91,
      "learning_rate": 4.138981225014733e-05,
      "loss": 0.0445,
      "step": 2011,
      "task_loss": 0.012897208333015442
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7558806967548358,
      "compression_loss": 0.0,
      "distillation_loss": 0.05814354866743088,
      "epoch": 1.91,
      "learning_rate": 4.1381764827103806e-05,
      "loss": 0.0544,
      "step": 2012,
      "task_loss": 0.0203425120562315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.755901332537123,
      "compression_loss": 0.0,
      "distillation_loss": 0.09011336416006088,
      "epoch": 1.91,
      "learning_rate": 4.13737144282338e-05,
      "loss": 0.0902,
      "step": 2013,
      "task_loss": 0.09071313589811325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7559219618838006,
      "compression_loss": 0.0,
      "distillation_loss": 0.055356744676828384,
      "epoch": 1.91,
      "learning_rate": 4.1365661054999715e-05,
      "loss": 0.0554,
      "step": 2014,
      "task_loss": 0.05614471435546875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7559425847958723,
      "compression_loss": 0.0,
      "distillation_loss": 0.1986648440361023,
      "epoch": 1.91,
      "learning_rate": 4.1357604708864475e-05,
      "loss": 0.2015,
      "step": 2015,
      "task_loss": 0.22709903120994568
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7559632012743418,
      "compression_loss": 0.0,
      "distillation_loss": 0.1105402335524559,
      "epoch": 1.91,
      "learning_rate": 4.1349545391291563e-05,
      "loss": 0.1051,
      "step": 2016,
      "task_loss": 0.05623643100261688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7559838113202129,
      "compression_loss": 0.0,
      "distillation_loss": 0.12236940860748291,
      "epoch": 1.92,
      "learning_rate": 4.1341483103745006e-05,
      "loss": 0.1174,
      "step": 2017,
      "task_loss": 0.07233867049217224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7560044149344889,
      "compression_loss": 0.0,
      "distillation_loss": 0.17869043350219727,
      "epoch": 1.92,
      "learning_rate": 4.133341784768933e-05,
      "loss": 0.1765,
      "step": 2018,
      "task_loss": 0.156291663646698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7560250121181739,
      "compression_loss": 0.0,
      "distillation_loss": 0.28349682688713074,
      "epoch": 1.92,
      "learning_rate": 4.132534962458962e-05,
      "loss": 0.2765,
      "step": 2019,
      "task_loss": 0.21306130290031433
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7560456028722713,
      "compression_loss": 0.0,
      "distillation_loss": 0.10640447586774826,
      "epoch": 1.92,
      "learning_rate": 4.131727843591155e-05,
      "loss": 0.1149,
      "step": 2020,
      "task_loss": 0.1911725401878357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.756066187197785,
      "compression_loss": 0.0,
      "distillation_loss": 0.13865335285663605,
      "epoch": 1.92,
      "learning_rate": 4.130920428312127e-05,
      "loss": 0.1432,
      "step": 2021,
      "task_loss": 0.18457633256912231
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7560867650957186,
      "compression_loss": 0.0,
      "distillation_loss": 0.06674148142337799,
      "epoch": 1.92,
      "learning_rate": 4.130112716768548e-05,
      "loss": 0.081,
      "step": 2022,
      "task_loss": 0.2092704176902771
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7561073365670756,
      "compression_loss": 0.0,
      "distillation_loss": 0.03668634220957756,
      "epoch": 1.92,
      "learning_rate": 4.129304709107143e-05,
      "loss": 0.0345,
      "step": 2023,
      "task_loss": 0.014327209442853928
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7561279016128599,
      "compression_loss": 0.0,
      "distillation_loss": 0.08399326354265213,
      "epoch": 1.92,
      "learning_rate": 4.128496405474691e-05,
      "loss": 0.0867,
      "step": 2024,
      "task_loss": 0.11154383420944214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7561484602340752,
      "compression_loss": 0.0,
      "distillation_loss": 0.16036614775657654,
      "epoch": 1.92,
      "learning_rate": 4.127687806018024e-05,
      "loss": 0.1671,
      "step": 2025,
      "task_loss": 0.22741574048995972
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7561690124317251,
      "compression_loss": 0.0,
      "distillation_loss": 0.02608412131667137,
      "epoch": 1.92,
      "learning_rate": 4.1268789108840275e-05,
      "loss": 0.0239,
      "step": 2026,
      "task_loss": 0.004486914724111557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7561895582068132,
      "compression_loss": 0.0,
      "distillation_loss": 0.04416799545288086,
      "epoch": 1.92,
      "learning_rate": 4.126069720219642e-05,
      "loss": 0.0411,
      "step": 2027,
      "task_loss": 0.013907143846154213
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7562100975603433,
      "compression_loss": 0.0,
      "distillation_loss": 0.24007698893547058,
      "epoch": 1.93,
      "learning_rate": 4.125260234171861e-05,
      "loss": 0.2428,
      "step": 2028,
      "task_loss": 0.2669823169708252
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7562306304933191,
      "compression_loss": 0.0,
      "distillation_loss": 0.19312305748462677,
      "epoch": 1.93,
      "learning_rate": 4.12445045288773e-05,
      "loss": 0.1849,
      "step": 2029,
      "task_loss": 0.11117222160100937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7562511570067442,
      "compression_loss": 0.0,
      "distillation_loss": 0.3301912546157837,
      "epoch": 1.93,
      "learning_rate": 4.123640376514353e-05,
      "loss": 0.316,
      "step": 2030,
      "task_loss": 0.18828773498535156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7562716771016224,
      "compression_loss": 0.0,
      "distillation_loss": 0.022117741405963898,
      "epoch": 1.93,
      "learning_rate": 4.12283000519888e-05,
      "loss": 0.0202,
      "step": 2031,
      "task_loss": 0.002740517258644104
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7562921907789572,
      "compression_loss": 0.0,
      "distillation_loss": 0.2477855086326599,
      "epoch": 1.93,
      "learning_rate": 4.122019339088522e-05,
      "loss": 0.2442,
      "step": 2032,
      "task_loss": 0.2123817652463913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7563126980397523,
      "compression_loss": 0.0,
      "distillation_loss": 0.06262822449207306,
      "epoch": 1.93,
      "learning_rate": 4.121208378330539e-05,
      "loss": 0.0674,
      "step": 2033,
      "task_loss": 0.11077867448329926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7563331988850116,
      "compression_loss": 0.0,
      "distillation_loss": 0.12019616365432739,
      "epoch": 1.93,
      "learning_rate": 4.120397123072246e-05,
      "loss": 0.1144,
      "step": 2034,
      "task_loss": 0.06211775913834572
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7563536933157385,
      "compression_loss": 0.0,
      "distillation_loss": 0.09517644345760345,
      "epoch": 1.93,
      "learning_rate": 4.119585573461012e-05,
      "loss": 0.087,
      "step": 2035,
      "task_loss": 0.013532513752579689
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.756374181332937,
      "compression_loss": 0.0,
      "distillation_loss": 0.1831834316253662,
      "epoch": 1.93,
      "learning_rate": 4.118773729644258e-05,
      "loss": 0.1754,
      "step": 2036,
      "task_loss": 0.10511328279972076
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7563946629376105,
      "compression_loss": 0.0,
      "distillation_loss": 0.17582698166370392,
      "epoch": 1.93,
      "learning_rate": 4.11796159176946e-05,
      "loss": 0.1714,
      "step": 2037,
      "task_loss": 0.13160312175750732
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7564151381307629,
      "compression_loss": 0.0,
      "distillation_loss": 0.19995486736297607,
      "epoch": 1.94,
      "learning_rate": 4.117149159984147e-05,
      "loss": 0.1948,
      "step": 2038,
      "task_loss": 0.14824716746807098
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7564356069133976,
      "compression_loss": 0.0,
      "distillation_loss": 0.33650684356689453,
      "epoch": 1.94,
      "learning_rate": 4.116336434435901e-05,
      "loss": 0.3217,
      "step": 2039,
      "task_loss": 0.1889341026544571
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7564560692865185,
      "compression_loss": 0.0,
      "distillation_loss": 0.22618404030799866,
      "epoch": 1.94,
      "learning_rate": 4.115523415272358e-05,
      "loss": 0.2265,
      "step": 2040,
      "task_loss": 0.2290044128894806
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7564765252511294,
      "compression_loss": 0.0,
      "distillation_loss": 0.08555667102336884,
      "epoch": 1.94,
      "learning_rate": 4.1147101026412046e-05,
      "loss": 0.0885,
      "step": 2041,
      "task_loss": 0.11483641713857651
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7564969748082336,
      "compression_loss": 0.0,
      "distillation_loss": 0.17309504747390747,
      "epoch": 1.94,
      "learning_rate": 4.1138964966901853e-05,
      "loss": 0.1651,
      "step": 2042,
      "task_loss": 0.09353820979595184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7565174179588352,
      "compression_loss": 0.0,
      "distillation_loss": 0.10289028286933899,
      "epoch": 1.94,
      "learning_rate": 4.113082597567095e-05,
      "loss": 0.1063,
      "step": 2043,
      "task_loss": 0.1365654170513153
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7565378547039375,
      "compression_loss": 0.0,
      "distillation_loss": 0.03402677923440933,
      "epoch": 1.94,
      "learning_rate": 4.112268405419782e-05,
      "loss": 0.0313,
      "step": 2044,
      "task_loss": 0.006699586287140846
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7565582850445445,
      "compression_loss": 0.0,
      "distillation_loss": 0.26117533445358276,
      "epoch": 1.94,
      "learning_rate": 4.1114539203961476e-05,
      "loss": 0.2482,
      "step": 2045,
      "task_loss": 0.13159014284610748
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7565787089816597,
      "compression_loss": 0.0,
      "distillation_loss": 0.1347775161266327,
      "epoch": 1.94,
      "learning_rate": 4.110639142644149e-05,
      "loss": 0.1373,
      "step": 2046,
      "task_loss": 0.16003359854221344
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7565991265162868,
      "compression_loss": 0.0,
      "distillation_loss": 0.2649126648902893,
      "epoch": 1.94,
      "learning_rate": 4.109824072311792e-05,
      "loss": 0.2607,
      "step": 2047,
      "task_loss": 0.2226344645023346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7566195376494295,
      "compression_loss": 0.0,
      "distillation_loss": 0.21198835968971252,
      "epoch": 1.94,
      "learning_rate": 4.10900870954714e-05,
      "loss": 0.2042,
      "step": 2048,
      "task_loss": 0.13424277305603027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7566399423820915,
      "compression_loss": 0.0,
      "distillation_loss": 0.16852621734142303,
      "epoch": 1.95,
      "learning_rate": 4.108193054498307e-05,
      "loss": 0.164,
      "step": 2049,
      "task_loss": 0.1233605444431305
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7566603407152765,
      "compression_loss": 0.0,
      "distillation_loss": 0.08903372287750244,
      "epoch": 1.95,
      "learning_rate": 4.10737710731346e-05,
      "loss": 0.0815,
      "step": 2050,
      "task_loss": 0.013720404356718063
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7566807326499883,
      "compression_loss": 0.0,
      "distillation_loss": 0.1362341046333313,
      "epoch": 1.95,
      "learning_rate": 4.106560868140821e-05,
      "loss": 0.1378,
      "step": 2051,
      "task_loss": 0.15231622755527496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7567011181872302,
      "compression_loss": 0.0,
      "distillation_loss": 0.12862615287303925,
      "epoch": 1.95,
      "learning_rate": 4.105744337128662e-05,
      "loss": 0.1334,
      "step": 2052,
      "task_loss": 0.17673933506011963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7567214973280062,
      "compression_loss": 0.0,
      "distillation_loss": 0.12414275854825974,
      "epoch": 1.95,
      "learning_rate": 4.104927514425312e-05,
      "loss": 0.1193,
      "step": 2053,
      "task_loss": 0.07598595321178436
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7567418700733199,
      "compression_loss": 0.0,
      "distillation_loss": 0.08933214098215103,
      "epoch": 1.95,
      "learning_rate": 4.104110400179148e-05,
      "loss": 0.0928,
      "step": 2054,
      "task_loss": 0.12388118356466293
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7567622364241751,
      "compression_loss": 0.0,
      "distillation_loss": 0.14208053052425385,
      "epoch": 1.95,
      "learning_rate": 4.103292994538605e-05,
      "loss": 0.1349,
      "step": 2055,
      "task_loss": 0.07043315470218658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7567825963815752,
      "compression_loss": 0.0,
      "distillation_loss": 0.26003506779670715,
      "epoch": 1.95,
      "learning_rate": 4.102475297652168e-05,
      "loss": 0.2527,
      "step": 2056,
      "task_loss": 0.18638324737548828
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7568029499465241,
      "compression_loss": 0.0,
      "distillation_loss": 0.13777679204940796,
      "epoch": 1.95,
      "learning_rate": 4.1016573096683765e-05,
      "loss": 0.1306,
      "step": 2057,
      "task_loss": 0.06621996313333511
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7568232971200255,
      "compression_loss": 0.0,
      "distillation_loss": 0.05839243531227112,
      "epoch": 1.95,
      "learning_rate": 4.10083903073582e-05,
      "loss": 0.053,
      "step": 2058,
      "task_loss": 0.004214171320199966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.756843637903083,
      "compression_loss": 0.0,
      "distillation_loss": 0.3341011106967926,
      "epoch": 1.96,
      "learning_rate": 4.1000204610031447e-05,
      "loss": 0.3197,
      "step": 2059,
      "task_loss": 0.18979643285274506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7568639722967002,
      "compression_loss": 0.0,
      "distillation_loss": 0.06705968081951141,
      "epoch": 1.96,
      "learning_rate": 4.0992016006190456e-05,
      "loss": 0.0707,
      "step": 2060,
      "task_loss": 0.10322168469429016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7568843003018809,
      "compression_loss": 0.0,
      "distillation_loss": 0.047776952385902405,
      "epoch": 1.96,
      "learning_rate": 4.0983824497322755e-05,
      "loss": 0.0474,
      "step": 2061,
      "task_loss": 0.04416225105524063
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7569046219196288,
      "compression_loss": 0.0,
      "distillation_loss": 0.10682905465364456,
      "epoch": 1.96,
      "learning_rate": 4.0975630084916344e-05,
      "loss": 0.1004,
      "step": 2062,
      "task_loss": 0.042584970593452454
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7569249371509476,
      "compression_loss": 0.0,
      "distillation_loss": 0.14277291297912598,
      "epoch": 1.96,
      "learning_rate": 4.096743277045979e-05,
      "loss": 0.1367,
      "step": 2063,
      "task_loss": 0.08212710171937943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7569452459968409,
      "compression_loss": 0.0,
      "distillation_loss": 0.20160989463329315,
      "epoch": 1.96,
      "learning_rate": 4.0959232555442174e-05,
      "loss": 0.194,
      "step": 2064,
      "task_loss": 0.12543994188308716
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7569655484583123,
      "compression_loss": 0.0,
      "distillation_loss": 0.12779472768306732,
      "epoch": 1.96,
      "learning_rate": 4.0951029441353104e-05,
      "loss": 0.1189,
      "step": 2065,
      "task_loss": 0.03901362046599388
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7569858445363656,
      "compression_loss": 0.0,
      "distillation_loss": 0.19314239919185638,
      "epoch": 1.96,
      "learning_rate": 4.094282342968271e-05,
      "loss": 0.1827,
      "step": 2066,
      "task_loss": 0.08886859565973282
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7570061342320046,
      "compression_loss": 0.0,
      "distillation_loss": 0.13019533455371857,
      "epoch": 1.96,
      "learning_rate": 4.093461452192167e-05,
      "loss": 0.1232,
      "step": 2067,
      "task_loss": 0.05998740345239639
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7570264175462328,
      "compression_loss": 0.0,
      "distillation_loss": 0.21118499338626862,
      "epoch": 1.96,
      "learning_rate": 4.092640271956115e-05,
      "loss": 0.2132,
      "step": 2068,
      "task_loss": 0.23096507787704468
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7570466944800539,
      "compression_loss": 0.0,
      "distillation_loss": 0.09960189461708069,
      "epoch": 1.96,
      "learning_rate": 4.091818802409288e-05,
      "loss": 0.0991,
      "step": 2069,
      "task_loss": 0.094699427485466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7570669650344717,
      "compression_loss": 0.0,
      "distillation_loss": 0.09009353071451187,
      "epoch": 1.97,
      "learning_rate": 4.0909970437009096e-05,
      "loss": 0.0879,
      "step": 2070,
      "task_loss": 0.06863107532262802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7570872292104897,
      "compression_loss": 0.0,
      "distillation_loss": 0.05240606516599655,
      "epoch": 1.97,
      "learning_rate": 4.0901749959802546e-05,
      "loss": 0.0545,
      "step": 2071,
      "task_loss": 0.07330939918756485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7571074870091117,
      "compression_loss": 0.0,
      "distillation_loss": 0.32454681396484375,
      "epoch": 1.97,
      "learning_rate": 4.0893526593966535e-05,
      "loss": 0.3176,
      "step": 2072,
      "task_loss": 0.25487715005874634
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7571277384313414,
      "compression_loss": 0.0,
      "distillation_loss": 0.1994623839855194,
      "epoch": 1.97,
      "learning_rate": 4.088530034099488e-05,
      "loss": 0.1925,
      "step": 2073,
      "task_loss": 0.13009962439537048
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7571479834781824,
      "compression_loss": 0.0,
      "distillation_loss": 0.043959006667137146,
      "epoch": 1.97,
      "learning_rate": 4.087707120238191e-05,
      "loss": 0.0632,
      "step": 2074,
      "task_loss": 0.2364317774772644
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7571682221506384,
      "compression_loss": 0.0,
      "distillation_loss": 0.1748720109462738,
      "epoch": 1.97,
      "learning_rate": 4.0868839179622495e-05,
      "loss": 0.1739,
      "step": 2075,
      "task_loss": 0.16508643329143524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7571884544497132,
      "compression_loss": 0.0,
      "distillation_loss": 0.09879680722951889,
      "epoch": 1.97,
      "learning_rate": 4.086060427421202e-05,
      "loss": 0.1112,
      "step": 2076,
      "task_loss": 0.2223973572254181
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7572086803764103,
      "compression_loss": 0.0,
      "distillation_loss": 0.13838914036750793,
      "epoch": 1.97,
      "learning_rate": 4.0852366487646384e-05,
      "loss": 0.1375,
      "step": 2077,
      "task_loss": 0.1297679990530014
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7572288999317336,
      "compression_loss": 0.0,
      "distillation_loss": 0.17635372281074524,
      "epoch": 1.97,
      "learning_rate": 4.084412582142204e-05,
      "loss": 0.1725,
      "step": 2078,
      "task_loss": 0.1377314180135727
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7572491131166865,
      "compression_loss": 0.0,
      "distillation_loss": 0.09490906447172165,
      "epoch": 1.97,
      "learning_rate": 4.083588227703593e-05,
      "loss": 0.1,
      "step": 2079,
      "task_loss": 0.1462094932794571
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.757269319932273,
      "compression_loss": 0.0,
      "distillation_loss": 0.05555427446961403,
      "epoch": 1.98,
      "learning_rate": 4.0827635855985534e-05,
      "loss": 0.0618,
      "step": 2080,
      "task_loss": 0.11752540618181229
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7572895203794965,
      "compression_loss": 0.0,
      "distillation_loss": 0.08329576253890991,
      "epoch": 1.98,
      "learning_rate": 4.081938655976886e-05,
      "loss": 0.0857,
      "step": 2081,
      "task_loss": 0.10723456740379333
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7573097144593608,
      "compression_loss": 0.0,
      "distillation_loss": 0.1329977810382843,
      "epoch": 1.98,
      "learning_rate": 4.0811134389884433e-05,
      "loss": 0.1404,
      "step": 2082,
      "task_loss": 0.2066575586795807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7573299021728697,
      "compression_loss": 0.0,
      "distillation_loss": 0.03028569370508194,
      "epoch": 1.98,
      "learning_rate": 4.08028793478313e-05,
      "loss": 0.0541,
      "step": 2083,
      "task_loss": 0.2682817280292511
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7573500835210266,
      "compression_loss": 0.0,
      "distillation_loss": 0.2019033133983612,
      "epoch": 1.98,
      "learning_rate": 4.0794621435109015e-05,
      "loss": 0.2013,
      "step": 2084,
      "task_loss": 0.19626502692699432
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7573702585048355,
      "compression_loss": 0.0,
      "distillation_loss": 0.04804403334856033,
      "epoch": 1.98,
      "learning_rate": 4.0786360653217684e-05,
      "loss": 0.049,
      "step": 2085,
      "task_loss": 0.05735541135072708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7573904271252998,
      "compression_loss": 0.0,
      "distillation_loss": 0.14371657371520996,
      "epoch": 1.98,
      "learning_rate": 4.0778097003657915e-05,
      "loss": 0.1367,
      "step": 2086,
      "task_loss": 0.07343168556690216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7574105893834234,
      "compression_loss": 0.0,
      "distillation_loss": 0.172747403383255,
      "epoch": 1.98,
      "learning_rate": 4.0769830487930835e-05,
      "loss": 0.1739,
      "step": 2087,
      "task_loss": 0.18447381258010864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7574307452802099,
      "compression_loss": 0.0,
      "distillation_loss": 0.18645472824573517,
      "epoch": 1.98,
      "learning_rate": 4.07615611075381e-05,
      "loss": 0.1843,
      "step": 2088,
      "task_loss": 0.16447779536247253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.757450894816663,
      "compression_loss": 0.0,
      "distillation_loss": 0.09333232790231705,
      "epoch": 1.98,
      "learning_rate": 4.075328886398188e-05,
      "loss": 0.102,
      "step": 2089,
      "task_loss": 0.17981281876564026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7574710379937862,
      "compression_loss": 0.0,
      "distillation_loss": 0.1386614441871643,
      "epoch": 1.98,
      "learning_rate": 4.074501375876487e-05,
      "loss": 0.1315,
      "step": 2090,
      "task_loss": 0.0667913407087326
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7574911748125835,
      "compression_loss": 0.0,
      "distillation_loss": 0.1604781299829483,
      "epoch": 1.99,
      "learning_rate": 4.073673579339028e-05,
      "loss": 0.155,
      "step": 2091,
      "task_loss": 0.10606161504983902
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7575113052740584,
      "compression_loss": 0.0,
      "distillation_loss": 0.2468898594379425,
      "epoch": 1.99,
      "learning_rate": 4.0728454969361854e-05,
      "loss": 0.25,
      "step": 2092,
      "task_loss": 0.2778300940990448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7575314293792146,
      "compression_loss": 0.0,
      "distillation_loss": 0.20150130987167358,
      "epoch": 1.99,
      "learning_rate": 4.0720171288183815e-05,
      "loss": 0.1922,
      "step": 2093,
      "task_loss": 0.10800500959157944
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7575515471290557,
      "compression_loss": 0.0,
      "distillation_loss": 0.1500411480665207,
      "epoch": 1.99,
      "learning_rate": 4.0711884751360964e-05,
      "loss": 0.1402,
      "step": 2094,
      "task_loss": 0.051413945853710175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7575716585245855,
      "compression_loss": 0.0,
      "distillation_loss": 0.052510011941194534,
      "epoch": 1.99,
      "learning_rate": 4.070359536039858e-05,
      "loss": 0.066,
      "step": 2095,
      "task_loss": 0.18779529631137848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7575917635668078,
      "compression_loss": 0.0,
      "distillation_loss": 0.2838015854358673,
      "epoch": 1.99,
      "learning_rate": 4.069530311680247e-05,
      "loss": 0.2762,
      "step": 2096,
      "task_loss": 0.20768359303474426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7576118622567259,
      "compression_loss": 0.0,
      "distillation_loss": 0.24412603676319122,
      "epoch": 1.99,
      "learning_rate": 4.068700802207895e-05,
      "loss": 0.2467,
      "step": 2097,
      "task_loss": 0.2701077461242676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7576319545953439,
      "compression_loss": 0.0,
      "distillation_loss": 0.18422961235046387,
      "epoch": 1.99,
      "learning_rate": 4.0678710077734885e-05,
      "loss": 0.1914,
      "step": 2098,
      "task_loss": 0.2559583783149719
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7576520405836652,
      "compression_loss": 0.0,
      "distillation_loss": 0.10283166915178299,
      "epoch": 1.99,
      "learning_rate": 4.0670409285277614e-05,
      "loss": 0.1053,
      "step": 2099,
      "task_loss": 0.12738347053527832
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7576721202226936,
      "compression_loss": 0.0,
      "distillation_loss": 0.24559719860553741,
      "epoch": 1.99,
      "learning_rate": 4.0662105646215034e-05,
      "loss": 0.2482,
      "step": 2100,
      "task_loss": 0.271691232919693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7576921935134328,
      "compression_loss": 0.0,
      "distillation_loss": 0.15227892994880676,
      "epoch": 2.0,
      "learning_rate": 4.065379916205554e-05,
      "loss": 0.1468,
      "step": 2101,
      "task_loss": 0.09752043336629868
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7577122604568864,
      "compression_loss": 0.0,
      "distillation_loss": 0.0908876582980156,
      "epoch": 2.0,
      "learning_rate": 4.0645489834308024e-05,
      "loss": 0.0848,
      "step": 2102,
      "task_loss": 0.02978145144879818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7577323210540582,
      "compression_loss": 0.0,
      "distillation_loss": 0.3263307809829712,
      "epoch": 2.0,
      "learning_rate": 4.063717766448194e-05,
      "loss": 0.3075,
      "step": 2103,
      "task_loss": 0.1381606012582779
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7577523753059517,
      "compression_loss": 0.0,
      "distillation_loss": 0.15767845511436462,
      "epoch": 2.0,
      "learning_rate": 4.062886265408722e-05,
      "loss": 0.1544,
      "step": 2104,
      "task_loss": 0.12524300813674927
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7577724232135707,
      "compression_loss": 0.0,
      "distillation_loss": 0.15870457887649536,
      "epoch": 2.0,
      "learning_rate": 4.062054480463433e-05,
      "loss": 0.1576,
      "step": 2105,
      "task_loss": 0.14793536067008972
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5726075510474254,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7330078136692348,
      "compression/magnitude_sparsity/target_sparsity_level": 0.757792464777919,
      "compression_loss": 0.0,
      "distillation_loss": 0.016080807894468307,
      "epoch": 2.0,
      "learning_rate": 4.0612224117634245e-05,
      "loss": 0.0148,
      "step": 2106,
      "task_loss": 0.003325019497424364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7578125,
      "compression_loss": 0.0,
      "distillation_loss": 0.8665308952331543,
      "epoch": 2.0,
      "learning_rate": 4.060390059459846e-05,
      "loss": 0.8038,
      "step": 2107,
      "task_loss": 0.23952460289001465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7578325288808176,
      "compression_loss": 0.0,
      "distillation_loss": 0.775876522064209,
      "epoch": 2.0,
      "learning_rate": 4.059557423703899e-05,
      "loss": 0.7176,
      "step": 2108,
      "task_loss": 0.1935485601425171
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7578525514213754,
      "compression_loss": 0.0,
      "distillation_loss": 0.8252198696136475,
      "epoch": 2.0,
      "learning_rate": 4.058724504646834e-05,
      "loss": 0.7657,
      "step": 2109,
      "task_loss": 0.230499267578125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7578725676226772,
      "compression_loss": 0.0,
      "distillation_loss": 0.6135385036468506,
      "epoch": 2.0,
      "learning_rate": 4.0578913024399564e-05,
      "loss": 0.5737,
      "step": 2110,
      "task_loss": 0.21483993530273438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7578925774857265,
      "compression_loss": 0.0,
      "distillation_loss": 0.5455840826034546,
      "epoch": 2.0,
      "learning_rate": 4.057057817234621e-05,
      "loss": 0.5087,
      "step": 2111,
      "task_loss": 0.17693614959716797
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.757912581011527,
      "compression_loss": 0.0,
      "distillation_loss": 0.43998754024505615,
      "epoch": 2.01,
      "learning_rate": 4.0562240491822334e-05,
      "loss": 0.4114,
      "step": 2112,
      "task_loss": 0.154291033744812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7579325782010825,
      "compression_loss": 0.0,
      "distillation_loss": 0.3841041922569275,
      "epoch": 2.01,
      "learning_rate": 4.055389998434253e-05,
      "loss": 0.3626,
      "step": 2113,
      "task_loss": 0.16882005333900452
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7579525690553967,
      "compression_loss": 0.0,
      "distillation_loss": 0.4492219090461731,
      "epoch": 2.01,
      "learning_rate": 4.054555665142189e-05,
      "loss": 0.4199,
      "step": 2114,
      "task_loss": 0.15565484762191772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7579725535754731,
      "compression_loss": 0.0,
      "distillation_loss": 0.3286392390727997,
      "epoch": 2.01,
      "learning_rate": 4.053721049457601e-05,
      "loss": 0.3141,
      "step": 2115,
      "task_loss": 0.1833527535200119
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7579925317623155,
      "compression_loss": 0.0,
      "distillation_loss": 0.2822880148887634,
      "epoch": 2.01,
      "learning_rate": 4.052886151532101e-05,
      "loss": 0.2741,
      "step": 2116,
      "task_loss": 0.2002839893102646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7580125036169276,
      "compression_loss": 0.0,
      "distillation_loss": 0.30642810463905334,
      "epoch": 2.01,
      "learning_rate": 4.0520509715173544e-05,
      "loss": 0.2994,
      "step": 2117,
      "task_loss": 0.235684335231781
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7580324691403132,
      "compression_loss": 0.0,
      "distillation_loss": 0.20797675848007202,
      "epoch": 2.01,
      "learning_rate": 4.051215509565073e-05,
      "loss": 0.2032,
      "step": 2118,
      "task_loss": 0.16037693619728088
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7580524283334756,
      "compression_loss": 0.0,
      "distillation_loss": 0.12344904243946075,
      "epoch": 2.01,
      "learning_rate": 4.050379765827024e-05,
      "loss": 0.1167,
      "step": 2119,
      "task_loss": 0.056173477321863174
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7580723811974189,
      "compression_loss": 0.0,
      "distillation_loss": 0.1896042823791504,
      "epoch": 2.01,
      "learning_rate": 4.0495437404550233e-05,
      "loss": 0.179,
      "step": 2120,
      "task_loss": 0.08344145119190216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7580923277331465,
      "compression_loss": 0.0,
      "distillation_loss": 0.16051128506660461,
      "epoch": 2.01,
      "learning_rate": 4.04870743360094e-05,
      "loss": 0.1519,
      "step": 2121,
      "task_loss": 0.07446075230836868
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7581122679416623,
      "compression_loss": 0.0,
      "distillation_loss": 0.37767481803894043,
      "epoch": 2.02,
      "learning_rate": 4.047870845416693e-05,
      "loss": 0.3772,
      "step": 2122,
      "task_loss": 0.3733970522880554
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7581322018239698,
      "compression_loss": 0.0,
      "distillation_loss": 0.21346813440322876,
      "epoch": 2.02,
      "learning_rate": 4.0470339760542506e-05,
      "loss": 0.2017,
      "step": 2123,
      "task_loss": 0.09529842436313629
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7581521293810727,
      "compression_loss": 0.0,
      "distillation_loss": 0.6465510129928589,
      "epoch": 2.02,
      "learning_rate": 4.0461968256656376e-05,
      "loss": 0.6242,
      "step": 2124,
      "task_loss": 0.42273980379104614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7581720506139747,
      "compression_loss": 0.0,
      "distillation_loss": 0.4466746747493744,
      "epoch": 2.02,
      "learning_rate": 4.045359394402925e-05,
      "loss": 0.4386,
      "step": 2125,
      "task_loss": 0.366089791059494
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7581919655236797,
      "compression_loss": 0.0,
      "distillation_loss": 0.05343036353588104,
      "epoch": 2.02,
      "learning_rate": 4.0445216824182344e-05,
      "loss": 0.0488,
      "step": 2126,
      "task_loss": 0.006774421781301498
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7582118741111911,
      "compression_loss": 0.0,
      "distillation_loss": 0.042159806936979294,
      "epoch": 2.02,
      "learning_rate": 4.043683689863742e-05,
      "loss": 0.0449,
      "step": 2127,
      "task_loss": 0.06956613063812256
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7582317763775126,
      "compression_loss": 0.0,
      "distillation_loss": 0.3928431272506714,
      "epoch": 2.02,
      "learning_rate": 4.042845416891673e-05,
      "loss": 0.3936,
      "step": 2128,
      "task_loss": 0.3999135494232178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7582516723236481,
      "compression_loss": 0.0,
      "distillation_loss": 0.22617900371551514,
      "epoch": 2.02,
      "learning_rate": 4.042006863654303e-05,
      "loss": 0.245,
      "step": 2129,
      "task_loss": 0.4148496389389038
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.758271561950601,
      "compression_loss": 0.0,
      "distillation_loss": 0.21618734300136566,
      "epoch": 2.02,
      "learning_rate": 4.041168030303961e-05,
      "loss": 0.207,
      "step": 2130,
      "task_loss": 0.12459345906972885
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7582914452593752,
      "compression_loss": 0.0,
      "distillation_loss": 0.398822546005249,
      "epoch": 2.02,
      "learning_rate": 4.0403289169930235e-05,
      "loss": 0.3847,
      "step": 2131,
      "task_loss": 0.2576183080673218
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7583113222509743,
      "compression_loss": 0.0,
      "distillation_loss": 0.16397595405578613,
      "epoch": 2.02,
      "learning_rate": 4.03948952387392e-05,
      "loss": 0.1636,
      "step": 2132,
      "task_loss": 0.15986183285713196
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.758331192926402,
      "compression_loss": 0.0,
      "distillation_loss": 0.19217489659786224,
      "epoch": 2.03,
      "learning_rate": 4.03864985109913e-05,
      "loss": 0.1823,
      "step": 2133,
      "task_loss": 0.09319883584976196
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.758351057286662,
      "compression_loss": 0.0,
      "distillation_loss": 0.16633236408233643,
      "epoch": 2.03,
      "learning_rate": 4.0378098988211845e-05,
      "loss": 0.1563,
      "step": 2134,
      "task_loss": 0.06592224538326263
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7583709153327578,
      "compression_loss": 0.0,
      "distillation_loss": 0.1656203418970108,
      "epoch": 2.03,
      "learning_rate": 4.036969667192665e-05,
      "loss": 0.1591,
      "step": 2135,
      "task_loss": 0.10005275160074234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7583907670656934,
      "compression_loss": 0.0,
      "distillation_loss": 0.1509263813495636,
      "epoch": 2.03,
      "learning_rate": 4.036129156366203e-05,
      "loss": 0.1424,
      "step": 2136,
      "task_loss": 0.0660940557718277
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7584106124864722,
      "compression_loss": 0.0,
      "distillation_loss": 0.1414584517478943,
      "epoch": 2.03,
      "learning_rate": 4.0352883664944816e-05,
      "loss": 0.137,
      "step": 2137,
      "task_loss": 0.09674646705389023
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7584304515960981,
      "compression_loss": 0.0,
      "distillation_loss": 0.10992178320884705,
      "epoch": 2.03,
      "learning_rate": 4.034447297730234e-05,
      "loss": 0.1093,
      "step": 2138,
      "task_loss": 0.10394454747438431
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7584502843955747,
      "compression_loss": 0.0,
      "distillation_loss": 0.2345104068517685,
      "epoch": 2.03,
      "learning_rate": 4.033605950226246e-05,
      "loss": 0.2226,
      "step": 2139,
      "task_loss": 0.115862175822258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7584701108859055,
      "compression_loss": 0.0,
      "distillation_loss": 0.15543803572654724,
      "epoch": 2.03,
      "learning_rate": 4.03276432413535e-05,
      "loss": 0.1514,
      "step": 2140,
      "task_loss": 0.11481384187936783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7584899310680945,
      "compression_loss": 0.0,
      "distillation_loss": 0.12512551248073578,
      "epoch": 2.03,
      "learning_rate": 4.0319224196104334e-05,
      "loss": 0.1191,
      "step": 2141,
      "task_loss": 0.06533493101596832
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7585097449431452,
      "compression_loss": 0.0,
      "distillation_loss": 0.09096786379814148,
      "epoch": 2.03,
      "learning_rate": 4.031080236804431e-05,
      "loss": 0.0902,
      "step": 2142,
      "task_loss": 0.08328302204608917
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7585295525120613,
      "compression_loss": 0.0,
      "distillation_loss": 0.10509166121482849,
      "epoch": 2.04,
      "learning_rate": 4.03023777587033e-05,
      "loss": 0.0993,
      "step": 2143,
      "task_loss": 0.04766244813799858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7585493537758464,
      "compression_loss": 0.0,
      "distillation_loss": 0.1337359994649887,
      "epoch": 2.04,
      "learning_rate": 4.029395036961168e-05,
      "loss": 0.1341,
      "step": 2144,
      "task_loss": 0.13710615038871765
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7585691487355044,
      "compression_loss": 0.0,
      "distillation_loss": 0.10559399425983429,
      "epoch": 2.04,
      "learning_rate": 4.028552020230031e-05,
      "loss": 0.0992,
      "step": 2145,
      "task_loss": 0.042120400816202164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7585889373920388,
      "compression_loss": 0.0,
      "distillation_loss": 0.04972688853740692,
      "epoch": 2.04,
      "learning_rate": 4.0277087258300575e-05,
      "loss": 0.0459,
      "step": 2146,
      "task_loss": 0.01133745163679123
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7586087197464534,
      "compression_loss": 0.0,
      "distillation_loss": 0.16767317056655884,
      "epoch": 2.04,
      "learning_rate": 4.0268651539144374e-05,
      "loss": 0.1856,
      "step": 2147,
      "task_loss": 0.3467387855052948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7586284957997518,
      "compression_loss": 0.0,
      "distillation_loss": 0.17021751403808594,
      "epoch": 2.04,
      "learning_rate": 4.026021304636408e-05,
      "loss": 0.1609,
      "step": 2148,
      "task_loss": 0.07744796574115753
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7586482655529376,
      "compression_loss": 0.0,
      "distillation_loss": 0.2166343480348587,
      "epoch": 2.04,
      "learning_rate": 4.0251771781492594e-05,
      "loss": 0.2155,
      "step": 2149,
      "task_loss": 0.20558899641036987
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7586680290070147,
      "compression_loss": 0.0,
      "distillation_loss": 0.23501603305339813,
      "epoch": 2.04,
      "learning_rate": 4.0243327746063315e-05,
      "loss": 0.2259,
      "step": 2150,
      "task_loss": 0.14403802156448364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7586877861629866,
      "compression_loss": 0.0,
      "distillation_loss": 0.21590901911258698,
      "epoch": 2.04,
      "learning_rate": 4.0234880941610134e-05,
      "loss": 0.2069,
      "step": 2151,
      "task_loss": 0.1256270408630371
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.758707537021857,
      "compression_loss": 0.0,
      "distillation_loss": 0.22788038849830627,
      "epoch": 2.04,
      "learning_rate": 4.022643136966746e-05,
      "loss": 0.2208,
      "step": 2152,
      "task_loss": 0.1574449986219406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7587272815846298,
      "compression_loss": 0.0,
      "distillation_loss": 0.34770405292510986,
      "epoch": 2.04,
      "learning_rate": 4.021797903177019e-05,
      "loss": 0.336,
      "step": 2153,
      "task_loss": 0.23103071749210358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7587470198523083,
      "compression_loss": 0.0,
      "distillation_loss": 0.13833436369895935,
      "epoch": 2.05,
      "learning_rate": 4.0209523929453744e-05,
      "loss": 0.1318,
      "step": 2154,
      "task_loss": 0.0730195865035057
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7587667518258966,
      "compression_loss": 0.0,
      "distillation_loss": 0.1359223574399948,
      "epoch": 2.05,
      "learning_rate": 4.0201066064254026e-05,
      "loss": 0.1326,
      "step": 2155,
      "task_loss": 0.10307220369577408
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.758786477506398,
      "compression_loss": 0.0,
      "distillation_loss": 0.19817477464675903,
      "epoch": 2.05,
      "learning_rate": 4.019260543770745e-05,
      "loss": 0.1921,
      "step": 2156,
      "task_loss": 0.13708311319351196
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7588061968948164,
      "compression_loss": 0.0,
      "distillation_loss": 0.1101415753364563,
      "epoch": 2.05,
      "learning_rate": 4.018414205135093e-05,
      "loss": 0.1024,
      "step": 2157,
      "task_loss": 0.03281085193157196
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7588259099921556,
      "compression_loss": 0.0,
      "distillation_loss": 0.2274709939956665,
      "epoch": 2.05,
      "learning_rate": 4.017567590672187e-05,
      "loss": 0.2163,
      "step": 2158,
      "task_loss": 0.11540090292692184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.758845616799419,
      "compression_loss": 0.0,
      "distillation_loss": 0.26632124185562134,
      "epoch": 2.05,
      "learning_rate": 4.01672070053582e-05,
      "loss": 0.2622,
      "step": 2159,
      "task_loss": 0.2254357933998108
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7588653173176103,
      "compression_loss": 0.0,
      "distillation_loss": 0.32481926679611206,
      "epoch": 2.05,
      "learning_rate": 4.015873534879833e-05,
      "loss": 0.3212,
      "step": 2160,
      "task_loss": 0.288277268409729
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7588850115477335,
      "compression_loss": 0.0,
      "distillation_loss": 0.2757709324359894,
      "epoch": 2.05,
      "learning_rate": 4.015026093858119e-05,
      "loss": 0.2797,
      "step": 2161,
      "task_loss": 0.3151787519454956
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7589046994907919,
      "compression_loss": 0.0,
      "distillation_loss": 0.07919088006019592,
      "epoch": 2.05,
      "learning_rate": 4.014178377624617e-05,
      "loss": 0.077,
      "step": 2162,
      "task_loss": 0.05737023800611496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7589243811477894,
      "compression_loss": 0.0,
      "distillation_loss": 0.13900256156921387,
      "epoch": 2.05,
      "learning_rate": 4.013330386333321e-05,
      "loss": 0.138,
      "step": 2163,
      "task_loss": 0.12849166989326477
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7589440565197296,
      "compression_loss": 0.0,
      "distillation_loss": 0.14013248682022095,
      "epoch": 2.06,
      "learning_rate": 4.012482120138272e-05,
      "loss": 0.1338,
      "step": 2164,
      "task_loss": 0.07710616290569305
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7589637256076164,
      "compression_loss": 0.0,
      "distillation_loss": 0.05423755198717117,
      "epoch": 2.06,
      "learning_rate": 4.011633579193561e-05,
      "loss": 0.0512,
      "step": 2165,
      "task_loss": 0.024237103760242462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7589833884124532,
      "compression_loss": 0.0,
      "distillation_loss": 0.20200000703334808,
      "epoch": 2.06,
      "learning_rate": 4.010784763653331e-05,
      "loss": 0.2011,
      "step": 2166,
      "task_loss": 0.19308245182037354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7590030449352436,
      "compression_loss": 0.0,
      "distillation_loss": 0.20488443970680237,
      "epoch": 2.06,
      "learning_rate": 4.0099356736717725e-05,
      "loss": 0.2045,
      "step": 2167,
      "task_loss": 0.20138651132583618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7590226951769916,
      "compression_loss": 0.0,
      "distillation_loss": 0.10349798947572708,
      "epoch": 2.06,
      "learning_rate": 4.0090863094031274e-05,
      "loss": 0.1,
      "step": 2168,
      "task_loss": 0.06868550926446915
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7590423391387008,
      "compression_loss": 0.0,
      "distillation_loss": 0.09170787781476974,
      "epoch": 2.06,
      "learning_rate": 4.008236671001686e-05,
      "loss": 0.0931,
      "step": 2169,
      "task_loss": 0.10580511391162872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7590619768213748,
      "compression_loss": 0.0,
      "distillation_loss": 0.15538711845874786,
      "epoch": 2.06,
      "learning_rate": 4.0073867586217895e-05,
      "loss": 0.1487,
      "step": 2170,
      "task_loss": 0.08846833556890488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7590816082260172,
      "compression_loss": 0.0,
      "distillation_loss": 0.1697424054145813,
      "epoch": 2.06,
      "learning_rate": 4.006536572417828e-05,
      "loss": 0.1626,
      "step": 2171,
      "task_loss": 0.09881134331226349
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7591012333536319,
      "compression_loss": 0.0,
      "distillation_loss": 0.13482381403446198,
      "epoch": 2.06,
      "learning_rate": 4.0056861125442435e-05,
      "loss": 0.1331,
      "step": 2172,
      "task_loss": 0.11754244565963745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7591208522052225,
      "compression_loss": 0.0,
      "distillation_loss": 0.29298990964889526,
      "epoch": 2.06,
      "learning_rate": 4.004835379155525e-05,
      "loss": 0.2928,
      "step": 2173,
      "task_loss": 0.291414350271225
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7591404647817925,
      "compression_loss": 0.0,
      "distillation_loss": 0.09788675606250763,
      "epoch": 2.06,
      "learning_rate": 4.003984372406212e-05,
      "loss": 0.0913,
      "step": 2174,
      "task_loss": 0.03239801526069641
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7591600710843459,
      "compression_loss": 0.0,
      "distillation_loss": 0.05986270681023598,
      "epoch": 2.07,
      "learning_rate": 4.003133092450895e-05,
      "loss": 0.0661,
      "step": 2175,
      "task_loss": 0.1226883977651596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7591796711138862,
      "compression_loss": 0.0,
      "distillation_loss": 0.038474880158901215,
      "epoch": 2.07,
      "learning_rate": 4.002281539444213e-05,
      "loss": 0.0356,
      "step": 2176,
      "task_loss": 0.00936584360897541
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.759199264871417,
      "compression_loss": 0.0,
      "distillation_loss": 0.11121018975973129,
      "epoch": 2.07,
      "learning_rate": 4.001429713540853e-05,
      "loss": 0.1127,
      "step": 2177,
      "task_loss": 0.12647242844104767
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7592188523579422,
      "compression_loss": 0.0,
      "distillation_loss": 0.11997029185295105,
      "epoch": 2.07,
      "learning_rate": 4.000577614895555e-05,
      "loss": 0.1197,
      "step": 2178,
      "task_loss": 0.11679819226264954
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7592384335744653,
      "compression_loss": 0.0,
      "distillation_loss": 0.2858255207538605,
      "epoch": 2.07,
      "learning_rate": 3.999725243663107e-05,
      "loss": 0.2816,
      "step": 2179,
      "task_loss": 0.24326691031455994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.75925800852199,
      "compression_loss": 0.0,
      "distillation_loss": 0.09184737503528595,
      "epoch": 2.07,
      "learning_rate": 3.9988725999983456e-05,
      "loss": 0.0963,
      "step": 2180,
      "task_loss": 0.13659027218818665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7592775772015201,
      "compression_loss": 0.0,
      "distillation_loss": 0.11890025436878204,
      "epoch": 2.07,
      "learning_rate": 3.998019684056158e-05,
      "loss": 0.121,
      "step": 2181,
      "task_loss": 0.13989683985710144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7592971396140592,
      "compression_loss": 0.0,
      "distillation_loss": 0.189500093460083,
      "epoch": 2.07,
      "learning_rate": 3.99716649599148e-05,
      "loss": 0.1928,
      "step": 2182,
      "task_loss": 0.22282639145851135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.759316695760611,
      "compression_loss": 0.0,
      "distillation_loss": 0.23618614673614502,
      "epoch": 2.07,
      "learning_rate": 3.996313035959297e-05,
      "loss": 0.2329,
      "step": 2183,
      "task_loss": 0.20297005772590637
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7593362456421792,
      "compression_loss": 0.0,
      "distillation_loss": 0.09011323750019073,
      "epoch": 2.07,
      "learning_rate": 3.995459304114645e-05,
      "loss": 0.0857,
      "step": 2184,
      "task_loss": 0.04603835195302963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7593557892597673,
      "compression_loss": 0.0,
      "distillation_loss": 0.14607930183410645,
      "epoch": 2.08,
      "learning_rate": 3.9946053006126086e-05,
      "loss": 0.1419,
      "step": 2185,
      "task_loss": 0.1042981743812561
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7593753266143792,
      "compression_loss": 0.0,
      "distillation_loss": 0.12250328063964844,
      "epoch": 2.08,
      "learning_rate": 3.993751025608321e-05,
      "loss": 0.1182,
      "step": 2186,
      "task_loss": 0.07953470945358276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7593948577070186,
      "compression_loss": 0.0,
      "distillation_loss": 0.1935318261384964,
      "epoch": 2.08,
      "learning_rate": 3.9928964792569655e-05,
      "loss": 0.1838,
      "step": 2187,
      "task_loss": 0.09669722616672516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7594143825386891,
      "compression_loss": 0.0,
      "distillation_loss": 0.07566056400537491,
      "epoch": 2.08,
      "learning_rate": 3.9920416617137745e-05,
      "loss": 0.0751,
      "step": 2188,
      "task_loss": 0.07045575231313705
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7594339011103943,
      "compression_loss": 0.0,
      "distillation_loss": 0.05975474417209625,
      "epoch": 2.08,
      "learning_rate": 3.9911865731340306e-05,
      "loss": 0.0558,
      "step": 2189,
      "task_loss": 0.020496994256973267
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7594534134231381,
      "compression_loss": 0.0,
      "distillation_loss": 0.1242765486240387,
      "epoch": 2.08,
      "learning_rate": 3.9903312136730634e-05,
      "loss": 0.1215,
      "step": 2190,
      "task_loss": 0.09669384360313416
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7594729194779241,
      "compression_loss": 0.0,
      "distillation_loss": 0.15067118406295776,
      "epoch": 2.08,
      "learning_rate": 3.989475583486254e-05,
      "loss": 0.1422,
      "step": 2191,
      "task_loss": 0.06594227254390717
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7594924192757557,
      "compression_loss": 0.0,
      "distillation_loss": 0.15141010284423828,
      "epoch": 2.08,
      "learning_rate": 3.988619682729032e-05,
      "loss": 0.1463,
      "step": 2192,
      "task_loss": 0.10027869790792465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.759511912817637,
      "compression_loss": 0.0,
      "distillation_loss": 0.08504438400268555,
      "epoch": 2.08,
      "learning_rate": 3.987763511556874e-05,
      "loss": 0.0815,
      "step": 2193,
      "task_loss": 0.0492318794131279
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7595314001045714,
      "compression_loss": 0.0,
      "distillation_loss": 0.03857031837105751,
      "epoch": 2.08,
      "learning_rate": 3.98690707012531e-05,
      "loss": 0.0354,
      "step": 2194,
      "task_loss": 0.006879139691591263
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7595508811375626,
      "compression_loss": 0.0,
      "distillation_loss": 0.12435384094715118,
      "epoch": 2.08,
      "learning_rate": 3.986050358589916e-05,
      "loss": 0.1221,
      "step": 2195,
      "task_loss": 0.10201095044612885
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7595703559176146,
      "compression_loss": 0.0,
      "distillation_loss": 0.13030223548412323,
      "epoch": 2.09,
      "learning_rate": 3.9851933771063166e-05,
      "loss": 0.1234,
      "step": 2196,
      "task_loss": 0.06090515851974487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7595898244457306,
      "compression_loss": 0.0,
      "distillation_loss": 0.21584641933441162,
      "epoch": 2.09,
      "learning_rate": 3.9843361258301876e-05,
      "loss": 0.204,
      "step": 2197,
      "task_loss": 0.09771460294723511
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7596092867229147,
      "compression_loss": 0.0,
      "distillation_loss": 0.09045465290546417,
      "epoch": 2.09,
      "learning_rate": 3.983478604917253e-05,
      "loss": 0.0834,
      "step": 2198,
      "task_loss": 0.01959792897105217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7596287427501703,
      "compression_loss": 0.0,
      "distillation_loss": 0.17018280923366547,
      "epoch": 2.09,
      "learning_rate": 3.9826208145232855e-05,
      "loss": 0.1639,
      "step": 2199,
      "task_loss": 0.10730506479740143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7596481925285012,
      "compression_loss": 0.0,
      "distillation_loss": 0.0529235415160656,
      "epoch": 2.09,
      "learning_rate": 3.981762754804107e-05,
      "loss": 0.0491,
      "step": 2200,
      "task_loss": 0.015178944915533066
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7596676360589112,
      "compression_loss": 0.0,
      "distillation_loss": 0.16582679748535156,
      "epoch": 2.09,
      "learning_rate": 3.980904425915586e-05,
      "loss": 0.157,
      "step": 2201,
      "task_loss": 0.07799357175827026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7596870733424037,
      "compression_loss": 0.0,
      "distillation_loss": 0.12228982895612717,
      "epoch": 2.09,
      "learning_rate": 3.9800458280136453e-05,
      "loss": 0.1221,
      "step": 2202,
      "task_loss": 0.12067455798387527
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7597065043799826,
      "compression_loss": 0.0,
      "distillation_loss": 0.28389573097229004,
      "epoch": 2.09,
      "learning_rate": 3.979186961254252e-05,
      "loss": 0.2762,
      "step": 2203,
      "task_loss": 0.2065102607011795
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7597259291726515,
      "compression_loss": 0.0,
      "distillation_loss": 0.04436230659484863,
      "epoch": 2.09,
      "learning_rate": 3.9783278257934233e-05,
      "loss": 0.0407,
      "step": 2204,
      "task_loss": 0.007897831499576569
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7597453477214141,
      "compression_loss": 0.0,
      "distillation_loss": 0.17565417289733887,
      "epoch": 2.09,
      "learning_rate": 3.977468421787225e-05,
      "loss": 0.169,
      "step": 2205,
      "task_loss": 0.10887274146080017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7597647600272741,
      "compression_loss": 0.0,
      "distillation_loss": 0.15632234513759613,
      "epoch": 2.09,
      "learning_rate": 3.976608749391773e-05,
      "loss": 0.1454,
      "step": 2206,
      "task_loss": 0.04727863147854805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7597841660912351,
      "compression_loss": 0.0,
      "distillation_loss": 0.2122887820005417,
      "epoch": 2.1,
      "learning_rate": 3.975748808763229e-05,
      "loss": 0.2159,
      "step": 2207,
      "task_loss": 0.24815846979618073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7598035659143009,
      "compression_loss": 0.0,
      "distillation_loss": 0.2334968000650406,
      "epoch": 2.1,
      "learning_rate": 3.974888600057808e-05,
      "loss": 0.2257,
      "step": 2208,
      "task_loss": 0.15524537861347198
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7598229594974751,
      "compression_loss": 0.0,
      "distillation_loss": 0.2860737442970276,
      "epoch": 2.1,
      "learning_rate": 3.974028123431769e-05,
      "loss": 0.2848,
      "step": 2209,
      "task_loss": 0.27306628227233887
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7598423468417614,
      "compression_loss": 0.0,
      "distillation_loss": 0.17595066130161285,
      "epoch": 2.1,
      "learning_rate": 3.973167379041421e-05,
      "loss": 0.186,
      "step": 2210,
      "task_loss": 0.2765977382659912
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7598617279481636,
      "compression_loss": 0.0,
      "distillation_loss": 0.049867670983076096,
      "epoch": 2.1,
      "learning_rate": 3.972306367043126e-05,
      "loss": 0.0474,
      "step": 2211,
      "task_loss": 0.02561107836663723
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7598811028176852,
      "compression_loss": 0.0,
      "distillation_loss": 0.13108716905117035,
      "epoch": 2.1,
      "learning_rate": 3.971445087593288e-05,
      "loss": 0.1296,
      "step": 2212,
      "task_loss": 0.11605143547058105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7599004714513299,
      "compression_loss": 0.0,
      "distillation_loss": 0.15945231914520264,
      "epoch": 2.1,
      "learning_rate": 3.970583540848363e-05,
      "loss": 0.1644,
      "step": 2213,
      "task_loss": 0.20883622765541077
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7599198338501015,
      "compression_loss": 0.0,
      "distillation_loss": 0.3196389973163605,
      "epoch": 2.1,
      "learning_rate": 3.969721726964856e-05,
      "loss": 0.3091,
      "step": 2214,
      "task_loss": 0.21398873627185822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7599391900150037,
      "compression_loss": 0.0,
      "distillation_loss": 0.30575287342071533,
      "epoch": 2.1,
      "learning_rate": 3.9688596460993176e-05,
      "loss": 0.29,
      "step": 2215,
      "task_loss": 0.1479155719280243
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7599585399470399,
      "compression_loss": 0.0,
      "distillation_loss": 0.08897826820611954,
      "epoch": 2.1,
      "learning_rate": 3.967997298408352e-05,
      "loss": 0.0882,
      "step": 2216,
      "task_loss": 0.08128625154495239
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7599778836472142,
      "compression_loss": 0.0,
      "distillation_loss": 0.0864962711930275,
      "epoch": 2.11,
      "learning_rate": 3.967134684048607e-05,
      "loss": 0.0896,
      "step": 2217,
      "task_loss": 0.11729071289300919
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7599972211165299,
      "compression_loss": 0.0,
      "distillation_loss": 0.23064328730106354,
      "epoch": 2.11,
      "learning_rate": 3.96627180317678e-05,
      "loss": 0.2177,
      "step": 2218,
      "task_loss": 0.10095411539077759
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.760016552355991,
      "compression_loss": 0.0,
      "distillation_loss": 0.13302667438983917,
      "epoch": 2.11,
      "learning_rate": 3.965408655949619e-05,
      "loss": 0.141,
      "step": 2219,
      "task_loss": 0.21288727223873138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7600358773666008,
      "compression_loss": 0.0,
      "distillation_loss": 0.27681732177734375,
      "epoch": 2.11,
      "learning_rate": 3.964545242523917e-05,
      "loss": 0.2902,
      "step": 2220,
      "task_loss": 0.41104960441589355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7600551961493635,
      "compression_loss": 0.0,
      "distillation_loss": 0.0774826779961586,
      "epoch": 2.11,
      "learning_rate": 3.9636815630565194e-05,
      "loss": 0.0723,
      "step": 2221,
      "task_loss": 0.026051845401525497
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7600745087052823,
      "compression_loss": 0.0,
      "distillation_loss": 0.0630701333284378,
      "epoch": 2.11,
      "learning_rate": 3.962817617704317e-05,
      "loss": 0.0647,
      "step": 2222,
      "task_loss": 0.07926364988088608
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7600938150353611,
      "compression_loss": 0.0,
      "distillation_loss": 0.22707828879356384,
      "epoch": 2.11,
      "learning_rate": 3.9619534066242485e-05,
      "loss": 0.2251,
      "step": 2223,
      "task_loss": 0.20687554776668549
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7601131151406035,
      "compression_loss": 0.0,
      "distillation_loss": 0.20871593058109283,
      "epoch": 2.11,
      "learning_rate": 3.961088929973303e-05,
      "loss": 0.2074,
      "step": 2224,
      "task_loss": 0.1955493688583374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7601324090220134,
      "compression_loss": 0.0,
      "distillation_loss": 0.08324743062257767,
      "epoch": 2.11,
      "learning_rate": 3.960224187908518e-05,
      "loss": 0.079,
      "step": 2225,
      "task_loss": 0.04101306200027466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7601516966805942,
      "compression_loss": 0.0,
      "distillation_loss": 0.22678276896476746,
      "epoch": 2.11,
      "learning_rate": 3.959359180586975e-05,
      "loss": 0.236,
      "step": 2226,
      "task_loss": 0.31925228238105774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7601709781173498,
      "compression_loss": 0.0,
      "distillation_loss": 0.15277154743671417,
      "epoch": 2.11,
      "learning_rate": 3.958493908165809e-05,
      "loss": 0.1434,
      "step": 2227,
      "task_loss": 0.05934750288724899
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7601902533332837,
      "compression_loss": 0.0,
      "distillation_loss": 0.13886742293834686,
      "epoch": 2.12,
      "learning_rate": 3.9576283708022e-05,
      "loss": 0.1342,
      "step": 2228,
      "task_loss": 0.09246313571929932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7602095223293998,
      "compression_loss": 0.0,
      "distillation_loss": 0.12244449555873871,
      "epoch": 2.12,
      "learning_rate": 3.956762568653378e-05,
      "loss": 0.1165,
      "step": 2229,
      "task_loss": 0.06263736635446548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7602287851067016,
      "compression_loss": 0.0,
      "distillation_loss": 0.11867455393075943,
      "epoch": 2.12,
      "learning_rate": 3.95589650187662e-05,
      "loss": 0.1288,
      "step": 2230,
      "task_loss": 0.21948017179965973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7602480416661928,
      "compression_loss": 0.0,
      "distillation_loss": 0.17664200067520142,
      "epoch": 2.12,
      "learning_rate": 3.95503017062925e-05,
      "loss": 0.1744,
      "step": 2231,
      "task_loss": 0.15381231904029846
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.760267292008877,
      "compression_loss": 0.0,
      "distillation_loss": 0.10297991335391998,
      "epoch": 2.12,
      "learning_rate": 3.954163575068643e-05,
      "loss": 0.0943,
      "step": 2232,
      "task_loss": 0.01588231697678566
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7602865361357581,
      "compression_loss": 0.0,
      "distillation_loss": 0.0686907023191452,
      "epoch": 2.12,
      "learning_rate": 3.953296715352218e-05,
      "loss": 0.0713,
      "step": 2233,
      "task_loss": 0.09433852881193161
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7603057740478397,
      "compression_loss": 0.0,
      "distillation_loss": 0.11666402965784073,
      "epoch": 2.12,
      "learning_rate": 3.952429591637446e-05,
      "loss": 0.1239,
      "step": 2234,
      "task_loss": 0.18875601887702942
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7603250057461255,
      "compression_loss": 0.0,
      "distillation_loss": 0.02542857825756073,
      "epoch": 2.12,
      "learning_rate": 3.951562204081845e-05,
      "loss": 0.0243,
      "step": 2235,
      "task_loss": 0.014179892838001251
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7603442312316191,
      "compression_loss": 0.0,
      "distillation_loss": 0.024139802902936935,
      "epoch": 2.12,
      "learning_rate": 3.950694552842977e-05,
      "loss": 0.0222,
      "step": 2236,
      "task_loss": 0.0047469362616539
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7603634505053243,
      "compression_loss": 0.0,
      "distillation_loss": 0.2176014482975006,
      "epoch": 2.12,
      "learning_rate": 3.949826638078457e-05,
      "loss": 0.2075,
      "step": 2237,
      "task_loss": 0.11611790210008621
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7603826635682446,
      "compression_loss": 0.0,
      "distillation_loss": 0.1716989278793335,
      "epoch": 2.13,
      "learning_rate": 3.948958459945946e-05,
      "loss": 0.1658,
      "step": 2238,
      "task_loss": 0.11308705061674118
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7604018704213837,
      "compression_loss": 0.0,
      "distillation_loss": 0.3528546392917633,
      "epoch": 2.13,
      "learning_rate": 3.948090018603153e-05,
      "loss": 0.3415,
      "step": 2239,
      "task_loss": 0.2390662580728531
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7604210710657455,
      "compression_loss": 0.0,
      "distillation_loss": 0.20263050496578217,
      "epoch": 2.13,
      "learning_rate": 3.947221314207834e-05,
      "loss": 0.2039,
      "step": 2240,
      "task_loss": 0.21529477834701538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7604402655023336,
      "compression_loss": 0.0,
      "distillation_loss": 0.11417360603809357,
      "epoch": 2.13,
      "learning_rate": 3.9463523469177935e-05,
      "loss": 0.1254,
      "step": 2241,
      "task_loss": 0.22623518109321594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7604594537321516,
      "compression_loss": 0.0,
      "distillation_loss": 0.05411672592163086,
      "epoch": 2.13,
      "learning_rate": 3.9454831168908824e-05,
      "loss": 0.0636,
      "step": 2242,
      "task_loss": 0.14903995394706726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7604786357562032,
      "compression_loss": 0.0,
      "distillation_loss": 0.09386488795280457,
      "epoch": 2.13,
      "learning_rate": 3.9446136242850025e-05,
      "loss": 0.1017,
      "step": 2243,
      "task_loss": 0.1717870980501175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7604978115754921,
      "compression_loss": 0.0,
      "distillation_loss": 0.056300774216651917,
      "epoch": 2.13,
      "learning_rate": 3.9437438692581e-05,
      "loss": 0.0646,
      "step": 2244,
      "task_loss": 0.1389637589454651
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.760516981191022,
      "compression_loss": 0.0,
      "distillation_loss": 0.0845174565911293,
      "epoch": 2.13,
      "learning_rate": 3.9428738519681704e-05,
      "loss": 0.0873,
      "step": 2245,
      "task_loss": 0.11270405352115631
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7605361446037966,
      "compression_loss": 0.0,
      "distillation_loss": 0.08052958548069,
      "epoch": 2.13,
      "learning_rate": 3.942003572573257e-05,
      "loss": 0.0794,
      "step": 2246,
      "task_loss": 0.06879500299692154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7605553018148195,
      "compression_loss": 0.0,
      "distillation_loss": 0.08333942294120789,
      "epoch": 2.13,
      "learning_rate": 3.94113303123145e-05,
      "loss": 0.0817,
      "step": 2247,
      "task_loss": 0.06727690994739532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7605744528250944,
      "compression_loss": 0.0,
      "distillation_loss": 0.11436310410499573,
      "epoch": 2.13,
      "learning_rate": 3.9402622281008874e-05,
      "loss": 0.1101,
      "step": 2248,
      "task_loss": 0.07128392904996872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7605935976356251,
      "compression_loss": 0.0,
      "distillation_loss": 0.2693362832069397,
      "epoch": 2.14,
      "learning_rate": 3.939391163339754e-05,
      "loss": 0.2577,
      "step": 2249,
      "task_loss": 0.1526666134595871
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7606127362474151,
      "compression_loss": 0.0,
      "distillation_loss": 0.07121552526950836,
      "epoch": 2.14,
      "learning_rate": 3.9385198371062845e-05,
      "loss": 0.0725,
      "step": 2250,
      "task_loss": 0.08450750261545181
    },
    {
      "epoch": 2.14,
      "eval_accuracy": 0.8818807339449541,
      "eval_loss": 0.48909926414489746,
      "eval_runtime": 18.0472,
      "eval_samples_per_second": 48.318,
      "eval_steps_per_second": 6.04,
      "step": 2250
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7606318686614681,
      "compression_loss": 0.0,
      "distillation_loss": 0.31588560342788696,
      "epoch": 2.14,
      "learning_rate": 3.937648249558758e-05,
      "loss": 0.3102,
      "step": 2251,
      "task_loss": 0.25923851132392883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7606509948787881,
      "compression_loss": 0.0,
      "distillation_loss": 0.0453728511929512,
      "epoch": 2.14,
      "learning_rate": 3.9367764008555034e-05,
      "loss": 0.0592,
      "step": 2252,
      "task_loss": 0.18407374620437622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7606701149003784,
      "compression_loss": 0.0,
      "distillation_loss": 0.16821657121181488,
      "epoch": 2.14,
      "learning_rate": 3.9359042911548955e-05,
      "loss": 0.1716,
      "step": 2253,
      "task_loss": 0.20164065062999725
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7606892287272428,
      "compression_loss": 0.0,
      "distillation_loss": 0.1436726152896881,
      "epoch": 2.14,
      "learning_rate": 3.935031920615358e-05,
      "loss": 0.1497,
      "step": 2254,
      "task_loss": 0.20403361320495605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.760708336360385,
      "compression_loss": 0.0,
      "distillation_loss": 0.40433016419410706,
      "epoch": 2.14,
      "learning_rate": 3.934159289395361e-05,
      "loss": 0.3865,
      "step": 2255,
      "task_loss": 0.22571320831775665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7607274378008088,
      "compression_loss": 0.0,
      "distillation_loss": 0.1408395618200302,
      "epoch": 2.14,
      "learning_rate": 3.9332863976534225e-05,
      "loss": 0.1363,
      "step": 2256,
      "task_loss": 0.09514202177524567
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7607465330495176,
      "compression_loss": 0.0,
      "distillation_loss": 0.18185870349407196,
      "epoch": 2.14,
      "learning_rate": 3.9324132455481064e-05,
      "loss": 0.173,
      "step": 2257,
      "task_loss": 0.09325390309095383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7607656221075153,
      "compression_loss": 0.0,
      "distillation_loss": 0.1712486296892166,
      "epoch": 2.14,
      "learning_rate": 3.931539833238026e-05,
      "loss": 0.1596,
      "step": 2258,
      "task_loss": 0.05442361161112785
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7607847049758056,
      "compression_loss": 0.0,
      "distillation_loss": 0.15352007746696472,
      "epoch": 2.15,
      "learning_rate": 3.930666160881841e-05,
      "loss": 0.1534,
      "step": 2259,
      "task_loss": 0.15225310623645782
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7608037816553921,
      "compression_loss": 0.0,
      "distillation_loss": 0.15940865874290466,
      "epoch": 2.15,
      "learning_rate": 3.9297922286382573e-05,
      "loss": 0.1447,
      "step": 2260,
      "task_loss": 0.012211665511131287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7608228521472785,
      "compression_loss": 0.0,
      "distillation_loss": 0.03598247095942497,
      "epoch": 2.15,
      "learning_rate": 3.928918036666029e-05,
      "loss": 0.0342,
      "step": 2261,
      "task_loss": 0.01777983456850052
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7608419164524685,
      "compression_loss": 0.0,
      "distillation_loss": 0.17762696743011475,
      "epoch": 2.15,
      "learning_rate": 3.928043585123957e-05,
      "loss": 0.1757,
      "step": 2262,
      "task_loss": 0.15850163996219635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7608609745719657,
      "compression_loss": 0.0,
      "distillation_loss": 0.0662430077791214,
      "epoch": 2.15,
      "learning_rate": 3.927168874170891e-05,
      "loss": 0.0623,
      "step": 2263,
      "task_loss": 0.026589026674628258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7608800265067739,
      "compression_loss": 0.0,
      "distillation_loss": 0.07591673731803894,
      "epoch": 2.15,
      "learning_rate": 3.926293903965726e-05,
      "loss": 0.0699,
      "step": 2264,
      "task_loss": 0.01550477184355259
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7608990722578967,
      "compression_loss": 0.0,
      "distillation_loss": 0.11433359235525131,
      "epoch": 2.15,
      "learning_rate": 3.925418674667405e-05,
      "loss": 0.1113,
      "step": 2265,
      "task_loss": 0.08434649556875229
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7609181118263377,
      "compression_loss": 0.0,
      "distillation_loss": 0.1352238655090332,
      "epoch": 2.15,
      "learning_rate": 3.924543186434915e-05,
      "loss": 0.1273,
      "step": 2266,
      "task_loss": 0.05604308471083641
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7609371452131007,
      "compression_loss": 0.0,
      "distillation_loss": 0.17770247161388397,
      "epoch": 2.15,
      "learning_rate": 3.923667439427295e-05,
      "loss": 0.1764,
      "step": 2267,
      "task_loss": 0.1651407778263092
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7609561724191896,
      "compression_loss": 0.0,
      "distillation_loss": 0.06021641939878464,
      "epoch": 2.15,
      "learning_rate": 3.922791433803629e-05,
      "loss": 0.0652,
      "step": 2268,
      "task_loss": 0.1097673624753952
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7609751934456076,
      "compression_loss": 0.0,
      "distillation_loss": 0.057901740074157715,
      "epoch": 2.15,
      "learning_rate": 3.921915169723046e-05,
      "loss": 0.0638,
      "step": 2269,
      "task_loss": 0.11656754463911057
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7609942082933588,
      "compression_loss": 0.0,
      "distillation_loss": 0.12669652700424194,
      "epoch": 2.16,
      "learning_rate": 3.921038647344725e-05,
      "loss": 0.1198,
      "step": 2270,
      "task_loss": 0.057524219155311584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7610132169634467,
      "compression_loss": 0.0,
      "distillation_loss": 0.11540260910987854,
      "epoch": 2.16,
      "learning_rate": 3.920161866827889e-05,
      "loss": 0.112,
      "step": 2271,
      "task_loss": 0.08180786669254303
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7610322194568749,
      "compression_loss": 0.0,
      "distillation_loss": 0.08194451034069061,
      "epoch": 2.16,
      "learning_rate": 3.9192848283318114e-05,
      "loss": 0.0761,
      "step": 2272,
      "task_loss": 0.023745771497488022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7610512157746473,
      "compression_loss": 0.0,
      "distillation_loss": 0.09653206914663315,
      "epoch": 2.16,
      "learning_rate": 3.91840753201581e-05,
      "loss": 0.0989,
      "step": 2273,
      "task_loss": 0.11974017322063446
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7610702059177674,
      "compression_loss": 0.0,
      "distillation_loss": 0.17683292925357819,
      "epoch": 2.16,
      "learning_rate": 3.917529978039247e-05,
      "loss": 0.1677,
      "step": 2274,
      "task_loss": 0.08567538112401962
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.761089189887239,
      "compression_loss": 0.0,
      "distillation_loss": 0.15673840045928955,
      "epoch": 2.16,
      "learning_rate": 3.9166521665615386e-05,
      "loss": 0.1618,
      "step": 2275,
      "task_loss": 0.20756980776786804
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7611081676840656,
      "compression_loss": 0.0,
      "distillation_loss": 0.16288813948631287,
      "epoch": 2.16,
      "learning_rate": 3.91577409774214e-05,
      "loss": 0.164,
      "step": 2276,
      "task_loss": 0.17406190931797028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7611271393092511,
      "compression_loss": 0.0,
      "distillation_loss": 0.09783841669559479,
      "epoch": 2.16,
      "learning_rate": 3.9148957717405596e-05,
      "loss": 0.0942,
      "step": 2277,
      "task_loss": 0.061522744596004486
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7611461047637991,
      "compression_loss": 0.0,
      "distillation_loss": 0.1405646800994873,
      "epoch": 2.16,
      "learning_rate": 3.914017188716347e-05,
      "loss": 0.1331,
      "step": 2278,
      "task_loss": 0.06549764424562454
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7611650640487132,
      "compression_loss": 0.0,
      "distillation_loss": 0.09124282747507095,
      "epoch": 2.16,
      "learning_rate": 3.913138348829102e-05,
      "loss": 0.0917,
      "step": 2279,
      "task_loss": 0.09613334387540817
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7611840171649973,
      "compression_loss": 0.0,
      "distillation_loss": 0.1638939529657364,
      "epoch": 2.17,
      "learning_rate": 3.91225925223847e-05,
      "loss": 0.161,
      "step": 2280,
      "task_loss": 0.13536801934242249
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7612029641136548,
      "compression_loss": 0.0,
      "distillation_loss": 0.0714389756321907,
      "epoch": 2.17,
      "learning_rate": 3.911379899104144e-05,
      "loss": 0.0746,
      "step": 2281,
      "task_loss": 0.10342606157064438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7612219048956895,
      "compression_loss": 0.0,
      "distillation_loss": 0.1365533024072647,
      "epoch": 2.17,
      "learning_rate": 3.910500289585862e-05,
      "loss": 0.142,
      "step": 2282,
      "task_loss": 0.19071495532989502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7612408395121052,
      "compression_loss": 0.0,
      "distillation_loss": 0.04966387897729874,
      "epoch": 2.17,
      "learning_rate": 3.90962042384341e-05,
      "loss": 0.0503,
      "step": 2283,
      "task_loss": 0.05590268224477768
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7612597679639055,
      "compression_loss": 0.0,
      "distillation_loss": 0.03692065179347992,
      "epoch": 2.17,
      "learning_rate": 3.908740302036618e-05,
      "loss": 0.0443,
      "step": 2284,
      "task_loss": 0.110334612429142
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.761278690252094,
      "compression_loss": 0.0,
      "distillation_loss": 0.10264438390731812,
      "epoch": 2.17,
      "learning_rate": 3.907859924325366e-05,
      "loss": 0.1084,
      "step": 2285,
      "task_loss": 0.16008785367012024
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7612976063776745,
      "compression_loss": 0.0,
      "distillation_loss": 0.11101731657981873,
      "epoch": 2.17,
      "learning_rate": 3.906979290869578e-05,
      "loss": 0.105,
      "step": 2286,
      "task_loss": 0.05065052583813667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7613165163416505,
      "compression_loss": 0.0,
      "distillation_loss": 0.14465484023094177,
      "epoch": 2.17,
      "learning_rate": 3.9060984018292267e-05,
      "loss": 0.1474,
      "step": 2287,
      "task_loss": 0.17257431149482727
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.761335420145026,
      "compression_loss": 0.0,
      "distillation_loss": 0.30154234170913696,
      "epoch": 2.17,
      "learning_rate": 3.905217257364328e-05,
      "loss": 0.2859,
      "step": 2288,
      "task_loss": 0.14557313919067383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7613543177888044,
      "compression_loss": 0.0,
      "distillation_loss": 0.09943203628063202,
      "epoch": 2.17,
      "learning_rate": 3.904335857634948e-05,
      "loss": 0.1025,
      "step": 2289,
      "task_loss": 0.12985821068286896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7613732092739895,
      "compression_loss": 0.0,
      "distillation_loss": 0.08901150524616241,
      "epoch": 2.17,
      "learning_rate": 3.9034542028011944e-05,
      "loss": 0.0838,
      "step": 2290,
      "task_loss": 0.03706664592027664
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.761392094601585,
      "compression_loss": 0.0,
      "distillation_loss": 0.01898990385234356,
      "epoch": 2.18,
      "learning_rate": 3.902572293023227e-05,
      "loss": 0.0269,
      "step": 2291,
      "task_loss": 0.09843745082616806
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7614109737725945,
      "compression_loss": 0.0,
      "distillation_loss": 0.04945681244134903,
      "epoch": 2.18,
      "learning_rate": 3.9016901284612474e-05,
      "loss": 0.046,
      "step": 2292,
      "task_loss": 0.015036612749099731
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7614298467880216,
      "compression_loss": 0.0,
      "distillation_loss": 0.07033106684684753,
      "epoch": 2.18,
      "learning_rate": 3.9008077092755055e-05,
      "loss": 0.0685,
      "step": 2293,
      "task_loss": 0.05228840187191963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7614487136488702,
      "compression_loss": 0.0,
      "distillation_loss": 0.02366098389029503,
      "epoch": 2.18,
      "learning_rate": 3.899925035626296e-05,
      "loss": 0.0333,
      "step": 2294,
      "task_loss": 0.1205047145485878
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.761467574356144,
      "compression_loss": 0.0,
      "distillation_loss": 0.11169403791427612,
      "epoch": 2.18,
      "learning_rate": 3.899042107673962e-05,
      "loss": 0.1119,
      "step": 2295,
      "task_loss": 0.11369898915290833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7614864289108464,
      "compression_loss": 0.0,
      "distillation_loss": 0.039377085864543915,
      "epoch": 2.18,
      "learning_rate": 3.898158925578893e-05,
      "loss": 0.0445,
      "step": 2296,
      "task_loss": 0.09109679609537125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7615052773139814,
      "compression_loss": 0.0,
      "distillation_loss": 0.32753410935401917,
      "epoch": 2.18,
      "learning_rate": 3.89727548950152e-05,
      "loss": 0.3187,
      "step": 2297,
      "task_loss": 0.23894454538822174
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7615241195665524,
      "compression_loss": 0.0,
      "distillation_loss": 0.13404498994350433,
      "epoch": 2.18,
      "learning_rate": 3.8963917996023245e-05,
      "loss": 0.1353,
      "step": 2298,
      "task_loss": 0.14633318781852722
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7615429556695633,
      "compression_loss": 0.0,
      "distillation_loss": 0.11489589512348175,
      "epoch": 2.18,
      "learning_rate": 3.8955078560418345e-05,
      "loss": 0.1183,
      "step": 2299,
      "task_loss": 0.14893510937690735
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7615617856240177,
      "compression_loss": 0.0,
      "distillation_loss": 0.02402464672923088,
      "epoch": 2.18,
      "learning_rate": 3.894623658980622e-05,
      "loss": 0.0309,
      "step": 2300,
      "task_loss": 0.09274609386920929
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7615806094309192,
      "compression_loss": 0.0,
      "distillation_loss": 0.12553593516349792,
      "epoch": 2.19,
      "learning_rate": 3.8937392085793036e-05,
      "loss": 0.1179,
      "step": 2301,
      "task_loss": 0.04914803430438042
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7615994270912717,
      "compression_loss": 0.0,
      "distillation_loss": 0.03810520097613335,
      "epoch": 2.19,
      "learning_rate": 3.892854504998546e-05,
      "loss": 0.0352,
      "step": 2302,
      "task_loss": 0.008859049528837204
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7616182386060786,
      "compression_loss": 0.0,
      "distillation_loss": 0.10612446069717407,
      "epoch": 2.19,
      "learning_rate": 3.891969548399061e-05,
      "loss": 0.1083,
      "step": 2303,
      "task_loss": 0.12744131684303284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7616370439763438,
      "compression_loss": 0.0,
      "distillation_loss": 0.08019978553056717,
      "epoch": 2.19,
      "learning_rate": 3.891084338941603e-05,
      "loss": 0.0734,
      "step": 2304,
      "task_loss": 0.012498440220952034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7616558432030709,
      "compression_loss": 0.0,
      "distillation_loss": 0.05349978804588318,
      "epoch": 2.19,
      "learning_rate": 3.8901988767869744e-05,
      "loss": 0.0752,
      "step": 2305,
      "task_loss": 0.2700839042663574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7616746362872636,
      "compression_loss": 0.0,
      "distillation_loss": 0.052905574440956116,
      "epoch": 2.19,
      "learning_rate": 3.8893131620960254e-05,
      "loss": 0.0546,
      "step": 2306,
      "task_loss": 0.06945595890283585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7616934232299255,
      "compression_loss": 0.0,
      "distillation_loss": 0.13802888989448547,
      "epoch": 2.19,
      "learning_rate": 3.88842719502965e-05,
      "loss": 0.1321,
      "step": 2307,
      "task_loss": 0.07853611558675766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7617122040320605,
      "compression_loss": 0.0,
      "distillation_loss": 0.06617145985364914,
      "epoch": 2.19,
      "learning_rate": 3.887540975748787e-05,
      "loss": 0.0631,
      "step": 2308,
      "task_loss": 0.03529322147369385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.761730978694672,
      "compression_loss": 0.0,
      "distillation_loss": 0.14270713925361633,
      "epoch": 2.19,
      "learning_rate": 3.8866545044144234e-05,
      "loss": 0.1358,
      "step": 2309,
      "task_loss": 0.07395662367343903
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7617497472187639,
      "compression_loss": 0.0,
      "distillation_loss": 0.12063492834568024,
      "epoch": 2.19,
      "learning_rate": 3.885767781187591e-05,
      "loss": 0.1182,
      "step": 2310,
      "task_loss": 0.09611006081104279
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7617685096053398,
      "compression_loss": 0.0,
      "distillation_loss": 0.15884768962860107,
      "epoch": 2.19,
      "learning_rate": 3.884880806229367e-05,
      "loss": 0.1541,
      "step": 2311,
      "task_loss": 0.1109381914138794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7617872658554034,
      "compression_loss": 0.0,
      "distillation_loss": 0.2637111246585846,
      "epoch": 2.2,
      "learning_rate": 3.883993579700875e-05,
      "loss": 0.2475,
      "step": 2312,
      "task_loss": 0.10145562142133713
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7618060159699582,
      "compression_loss": 0.0,
      "distillation_loss": 0.09027013182640076,
      "epoch": 2.2,
      "learning_rate": 3.883106101763285e-05,
      "loss": 0.0899,
      "step": 2313,
      "task_loss": 0.08697813749313354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7618247599500082,
      "compression_loss": 0.0,
      "distillation_loss": 0.10050283372402191,
      "epoch": 2.2,
      "learning_rate": 3.882218372577809e-05,
      "loss": 0.0967,
      "step": 2314,
      "task_loss": 0.062062062323093414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7618434977965569,
      "compression_loss": 0.0,
      "distillation_loss": 0.21101830899715424,
      "epoch": 2.2,
      "learning_rate": 3.881330392305709e-05,
      "loss": 0.2235,
      "step": 2315,
      "task_loss": 0.3360675871372223
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.761862229510608,
      "compression_loss": 0.0,
      "distillation_loss": 0.20589599013328552,
      "epoch": 2.2,
      "learning_rate": 3.8804421611082916e-05,
      "loss": 0.1933,
      "step": 2316,
      "task_loss": 0.07991321384906769
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7618809550931652,
      "compression_loss": 0.0,
      "distillation_loss": 0.1543605923652649,
      "epoch": 2.2,
      "learning_rate": 3.8795536791469066e-05,
      "loss": 0.1454,
      "step": 2317,
      "task_loss": 0.06446463614702225
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7618996745452322,
      "compression_loss": 0.0,
      "distillation_loss": 0.09525361657142639,
      "epoch": 2.2,
      "learning_rate": 3.8786649465829516e-05,
      "loss": 0.0921,
      "step": 2318,
      "task_loss": 0.0634106770157814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7619183878678126,
      "compression_loss": 0.0,
      "distillation_loss": 0.041672758758068085,
      "epoch": 2.2,
      "learning_rate": 3.8777759635778696e-05,
      "loss": 0.0385,
      "step": 2319,
      "task_loss": 0.009624077007174492
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7619370950619102,
      "compression_loss": 0.0,
      "distillation_loss": 0.13709776103496552,
      "epoch": 2.2,
      "learning_rate": 3.876886730293149e-05,
      "loss": 0.1341,
      "step": 2320,
      "task_loss": 0.10736387968063354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7619557961285286,
      "compression_loss": 0.0,
      "distillation_loss": 0.0763949453830719,
      "epoch": 2.2,
      "learning_rate": 3.8759972468903215e-05,
      "loss": 0.0731,
      "step": 2321,
      "task_loss": 0.04355807602405548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7619744910686715,
      "compression_loss": 0.0,
      "distillation_loss": 0.08063163608312607,
      "epoch": 2.21,
      "learning_rate": 3.875107513530968e-05,
      "loss": 0.0841,
      "step": 2322,
      "task_loss": 0.11514770984649658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7619931798833426,
      "compression_loss": 0.0,
      "distillation_loss": 0.05136069282889366,
      "epoch": 2.21,
      "learning_rate": 3.874217530376711e-05,
      "loss": 0.0527,
      "step": 2323,
      "task_loss": 0.06518974155187607
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7620118625735456,
      "compression_loss": 0.0,
      "distillation_loss": 0.045473020523786545,
      "epoch": 2.21,
      "learning_rate": 3.873327297589223e-05,
      "loss": 0.0417,
      "step": 2324,
      "task_loss": 0.007644519209861755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7620305391402841,
      "compression_loss": 0.0,
      "distillation_loss": 0.033313095569610596,
      "epoch": 2.21,
      "learning_rate": 3.8724368153302166e-05,
      "loss": 0.0488,
      "step": 2325,
      "task_loss": 0.18855777382850647
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7620492095845618,
      "compression_loss": 0.0,
      "distillation_loss": 0.23782016336917877,
      "epoch": 2.21,
      "learning_rate": 3.871546083761453e-05,
      "loss": 0.2333,
      "step": 2326,
      "task_loss": 0.19272853434085846
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7620678739073825,
      "compression_loss": 0.0,
      "distillation_loss": 0.10806293785572052,
      "epoch": 2.21,
      "learning_rate": 3.870655103044738e-05,
      "loss": 0.1051,
      "step": 2327,
      "task_loss": 0.07805277407169342
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7620865321097497,
      "compression_loss": 0.0,
      "distillation_loss": 0.1450902819633484,
      "epoch": 2.21,
      "learning_rate": 3.8697638733419216e-05,
      "loss": 0.1578,
      "step": 2328,
      "task_loss": 0.2723790109157562
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7621051841926673,
      "compression_loss": 0.0,
      "distillation_loss": 0.09044960141181946,
      "epoch": 2.21,
      "learning_rate": 3.8688723948149014e-05,
      "loss": 0.0879,
      "step": 2329,
      "task_loss": 0.0644555538892746
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7621238301571387,
      "compression_loss": 0.0,
      "distillation_loss": 0.036163873970508575,
      "epoch": 2.21,
      "learning_rate": 3.867980667625618e-05,
      "loss": 0.0502,
      "step": 2330,
      "task_loss": 0.17629528045654297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7621424700041679,
      "compression_loss": 0.0,
      "distillation_loss": 0.08255277574062347,
      "epoch": 2.21,
      "learning_rate": 3.867088691936058e-05,
      "loss": 0.0778,
      "step": 2331,
      "task_loss": 0.03511197119951248
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7621611037347583,
      "compression_loss": 0.0,
      "distillation_loss": 0.1695278137922287,
      "epoch": 2.21,
      "learning_rate": 3.8661964679082535e-05,
      "loss": 0.1618,
      "step": 2332,
      "task_loss": 0.09199361503124237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7621797313499138,
      "compression_loss": 0.0,
      "distillation_loss": 0.052597202360630035,
      "epoch": 2.22,
      "learning_rate": 3.8653039957042806e-05,
      "loss": 0.0562,
      "step": 2333,
      "task_loss": 0.08904554694890976
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.762198352850638,
      "compression_loss": 0.0,
      "distillation_loss": 0.154799684882164,
      "epoch": 2.22,
      "learning_rate": 3.8644112754862614e-05,
      "loss": 0.1498,
      "step": 2334,
      "task_loss": 0.10432252287864685
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7622169682379345,
      "compression_loss": 0.0,
      "distillation_loss": 0.06855852156877518,
      "epoch": 2.22,
      "learning_rate": 3.8635183074163636e-05,
      "loss": 0.0654,
      "step": 2335,
      "task_loss": 0.03740303963422775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7622355775128071,
      "compression_loss": 0.0,
      "distillation_loss": 0.031943485140800476,
      "epoch": 2.22,
      "learning_rate": 3.862625091656797e-05,
      "loss": 0.039,
      "step": 2336,
      "task_loss": 0.1027441918849945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7622541806762594,
      "compression_loss": 0.0,
      "distillation_loss": 0.08677913248538971,
      "epoch": 2.22,
      "learning_rate": 3.861731628369822e-05,
      "loss": 0.0888,
      "step": 2337,
      "task_loss": 0.10726401954889297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7622727777292951,
      "compression_loss": 0.0,
      "distillation_loss": 0.03856203332543373,
      "epoch": 2.22,
      "learning_rate": 3.8608379177177375e-05,
      "loss": 0.0354,
      "step": 2338,
      "task_loss": 0.0071256812661886215
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7622913686729179,
      "compression_loss": 0.0,
      "distillation_loss": 0.04021540284156799,
      "epoch": 2.22,
      "learning_rate": 3.8599439598628916e-05,
      "loss": 0.0368,
      "step": 2339,
      "task_loss": 0.005587218329310417
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7623099535081316,
      "compression_loss": 0.0,
      "distillation_loss": 0.035530779510736465,
      "epoch": 2.22,
      "learning_rate": 3.8590497549676753e-05,
      "loss": 0.0384,
      "step": 2340,
      "task_loss": 0.06408775597810745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7623285322359397,
      "compression_loss": 0.0,
      "distillation_loss": 0.16440139710903168,
      "epoch": 2.22,
      "learning_rate": 3.858155303194526e-05,
      "loss": 0.1682,
      "step": 2341,
      "task_loss": 0.20209884643554688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7623471048573459,
      "compression_loss": 0.0,
      "distillation_loss": 0.09271198511123657,
      "epoch": 2.22,
      "learning_rate": 3.8572606047059254e-05,
      "loss": 0.0869,
      "step": 2342,
      "task_loss": 0.03472534567117691
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.762365671373354,
      "compression_loss": 0.0,
      "distillation_loss": 0.22541561722755432,
      "epoch": 2.23,
      "learning_rate": 3.856365659664399e-05,
      "loss": 0.2111,
      "step": 2343,
      "task_loss": 0.08197568356990814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7623842317849675,
      "compression_loss": 0.0,
      "distillation_loss": 0.1988585889339447,
      "epoch": 2.23,
      "learning_rate": 3.855470468232518e-05,
      "loss": 0.2026,
      "step": 2344,
      "task_loss": 0.23670420050621033
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7624027860931902,
      "compression_loss": 0.0,
      "distillation_loss": 0.08545896410942078,
      "epoch": 2.23,
      "learning_rate": 3.854575030572898e-05,
      "loss": 0.0809,
      "step": 2345,
      "task_loss": 0.040093451738357544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7624213342990259,
      "compression_loss": 0.0,
      "distillation_loss": 0.11442442238330841,
      "epoch": 2.23,
      "learning_rate": 3.853679346848201e-05,
      "loss": 0.1172,
      "step": 2346,
      "task_loss": 0.14235654473304749
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7624398764034781,
      "compression_loss": 0.0,
      "distillation_loss": 0.08486449718475342,
      "epoch": 2.23,
      "learning_rate": 3.8527834172211306e-05,
      "loss": 0.0992,
      "step": 2347,
      "task_loss": 0.22856228053569794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7624584124075505,
      "compression_loss": 0.0,
      "distillation_loss": 0.13581222295761108,
      "epoch": 2.23,
      "learning_rate": 3.851887241854438e-05,
      "loss": 0.129,
      "step": 2348,
      "task_loss": 0.06761342287063599
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7624769423122468,
      "compression_loss": 0.0,
      "distillation_loss": 0.3619718551635742,
      "epoch": 2.23,
      "learning_rate": 3.850990820910917e-05,
      "loss": 0.3528,
      "step": 2349,
      "task_loss": 0.27037566900253296
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7624954661185708,
      "compression_loss": 0.0,
      "distillation_loss": 0.1374235451221466,
      "epoch": 2.23,
      "learning_rate": 3.8500941545534065e-05,
      "loss": 0.137,
      "step": 2350,
      "task_loss": 0.13285070657730103
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.762513983827526,
      "compression_loss": 0.0,
      "distillation_loss": 0.03058495558798313,
      "epoch": 2.23,
      "learning_rate": 3.849197242944791e-05,
      "loss": 0.0281,
      "step": 2351,
      "task_loss": 0.0058120060712099075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7625324954401163,
      "compression_loss": 0.0,
      "distillation_loss": 0.26459288597106934,
      "epoch": 2.23,
      "learning_rate": 3.8483000862479986e-05,
      "loss": 0.2585,
      "step": 2352,
      "task_loss": 0.2032066434621811
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7625510009573451,
      "compression_loss": 0.0,
      "distillation_loss": 0.11497370898723602,
      "epoch": 2.23,
      "learning_rate": 3.8474026846260015e-05,
      "loss": 0.1074,
      "step": 2353,
      "task_loss": 0.039116568863391876
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7625695003802162,
      "compression_loss": 0.0,
      "distillation_loss": 0.10177917033433914,
      "epoch": 2.24,
      "learning_rate": 3.846505038241818e-05,
      "loss": 0.098,
      "step": 2354,
      "task_loss": 0.0641433447599411
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7625879937097334,
      "compression_loss": 0.0,
      "distillation_loss": 0.1023842841386795,
      "epoch": 2.24,
      "learning_rate": 3.84560714725851e-05,
      "loss": 0.1054,
      "step": 2355,
      "task_loss": 0.13256244361400604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7626064809469003,
      "compression_loss": 0.0,
      "distillation_loss": 0.21765132248401642,
      "epoch": 2.24,
      "learning_rate": 3.8447090118391814e-05,
      "loss": 0.2109,
      "step": 2356,
      "task_loss": 0.14984160661697388
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7626249620927206,
      "compression_loss": 0.0,
      "distillation_loss": 0.10686811059713364,
      "epoch": 2.24,
      "learning_rate": 3.8438106321469864e-05,
      "loss": 0.1046,
      "step": 2357,
      "task_loss": 0.08399415016174316
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7626434371481979,
      "compression_loss": 0.0,
      "distillation_loss": 0.06657497584819794,
      "epoch": 2.24,
      "learning_rate": 3.842912008345117e-05,
      "loss": 0.0629,
      "step": 2358,
      "task_loss": 0.029489833861589432
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.762661906114336,
      "compression_loss": 0.0,
      "distillation_loss": 0.17578309774398804,
      "epoch": 2.24,
      "learning_rate": 3.842013140596815e-05,
      "loss": 0.1783,
      "step": 2359,
      "task_loss": 0.20061953365802765
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7626803689921385,
      "compression_loss": 0.0,
      "distillation_loss": 0.08999986201524734,
      "epoch": 2.24,
      "learning_rate": 3.841114029065362e-05,
      "loss": 0.1037,
      "step": 2360,
      "task_loss": 0.22721917927265167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7626988257826091,
      "compression_loss": 0.0,
      "distillation_loss": 0.18849384784698486,
      "epoch": 2.24,
      "learning_rate": 3.8402146739140874e-05,
      "loss": 0.1817,
      "step": 2361,
      "task_loss": 0.12036500871181488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7627172764867515,
      "compression_loss": 0.0,
      "distillation_loss": 0.03334838151931763,
      "epoch": 2.24,
      "learning_rate": 3.8393150753063614e-05,
      "loss": 0.0326,
      "step": 2362,
      "task_loss": 0.025991789996623993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7627357211055693,
      "compression_loss": 0.0,
      "distillation_loss": 0.09625150263309479,
      "epoch": 2.24,
      "learning_rate": 3.838415233405603e-05,
      "loss": 0.1046,
      "step": 2363,
      "task_loss": 0.17937156558036804
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7627541596400663,
      "compression_loss": 0.0,
      "distillation_loss": 0.07551825791597366,
      "epoch": 2.25,
      "learning_rate": 3.837515148375271e-05,
      "loss": 0.0853,
      "step": 2364,
      "task_loss": 0.17311903834342957
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7627725920912461,
      "compression_loss": 0.0,
      "distillation_loss": 0.06278804689645767,
      "epoch": 2.25,
      "learning_rate": 3.836614820378871e-05,
      "loss": 0.0712,
      "step": 2365,
      "task_loss": 0.14707757532596588
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7627910184601125,
      "compression_loss": 0.0,
      "distillation_loss": 0.045216597616672516,
      "epoch": 2.25,
      "learning_rate": 3.835714249579952e-05,
      "loss": 0.0581,
      "step": 2366,
      "task_loss": 0.1742611825466156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.762809438747669,
      "compression_loss": 0.0,
      "distillation_loss": 0.28981393575668335,
      "epoch": 2.25,
      "learning_rate": 3.8348134361421064e-05,
      "loss": 0.2989,
      "step": 2367,
      "task_loss": 0.38077259063720703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7628278529549195,
      "compression_loss": 0.0,
      "distillation_loss": 0.0407429076731205,
      "epoch": 2.25,
      "learning_rate": 3.8339123802289716e-05,
      "loss": 0.037,
      "step": 2368,
      "task_loss": 0.0035578403621912003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7628462610828675,
      "compression_loss": 0.0,
      "distillation_loss": 0.08070795238018036,
      "epoch": 2.25,
      "learning_rate": 3.8330110820042285e-05,
      "loss": 0.0862,
      "step": 2369,
      "task_loss": 0.1353619545698166
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7628646631325168,
      "compression_loss": 0.0,
      "distillation_loss": 0.05324525758624077,
      "epoch": 2.25,
      "learning_rate": 3.8321095416316024e-05,
      "loss": 0.0695,
      "step": 2370,
      "task_loss": 0.21613724529743195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.762883059104871,
      "compression_loss": 0.0,
      "distillation_loss": 0.04425017535686493,
      "epoch": 2.25,
      "learning_rate": 3.831207759274863e-05,
      "loss": 0.0407,
      "step": 2371,
      "task_loss": 0.008313026279211044
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7629014490009337,
      "compression_loss": 0.0,
      "distillation_loss": 0.07859969139099121,
      "epoch": 2.25,
      "learning_rate": 3.8303057350978224e-05,
      "loss": 0.1042,
      "step": 2372,
      "task_loss": 0.3349103033542633
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7629198328217088,
      "compression_loss": 0.0,
      "distillation_loss": 0.09541267156600952,
      "epoch": 2.25,
      "learning_rate": 3.829403469264339e-05,
      "loss": 0.0935,
      "step": 2373,
      "task_loss": 0.07653743773698807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7629382105681998,
      "compression_loss": 0.0,
      "distillation_loss": 0.10529537498950958,
      "epoch": 2.25,
      "learning_rate": 3.828500961938313e-05,
      "loss": 0.0988,
      "step": 2374,
      "task_loss": 0.040317974984645844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7629565822414105,
      "compression_loss": 0.0,
      "distillation_loss": 0.050773873925209045,
      "epoch": 2.26,
      "learning_rate": 3.827598213283688e-05,
      "loss": 0.0477,
      "step": 2375,
      "task_loss": 0.019673151895403862
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7629749478423445,
      "compression_loss": 0.0,
      "distillation_loss": 0.06847432255744934,
      "epoch": 2.26,
      "learning_rate": 3.8266952234644545e-05,
      "loss": 0.0684,
      "step": 2376,
      "task_loss": 0.06723335385322571
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7629933073720057,
      "compression_loss": 0.0,
      "distillation_loss": 0.1094050258398056,
      "epoch": 2.26,
      "learning_rate": 3.825791992644644e-05,
      "loss": 0.1155,
      "step": 2377,
      "task_loss": 0.17028164863586426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7630116608313974,
      "compression_loss": 0.0,
      "distillation_loss": 0.04125671088695526,
      "epoch": 2.26,
      "learning_rate": 3.824888520988333e-05,
      "loss": 0.038,
      "step": 2378,
      "task_loss": 0.008307870477437973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7630300082215236,
      "compression_loss": 0.0,
      "distillation_loss": 0.0585186630487442,
      "epoch": 2.26,
      "learning_rate": 3.823984808659641e-05,
      "loss": 0.0588,
      "step": 2379,
      "task_loss": 0.06130526214838028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7630483495433877,
      "compression_loss": 0.0,
      "distillation_loss": 0.03627898544073105,
      "epoch": 2.26,
      "learning_rate": 3.8230808558227335e-05,
      "loss": 0.0331,
      "step": 2380,
      "task_loss": 0.004686055704951286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7630666847979937,
      "compression_loss": 0.0,
      "distillation_loss": 0.04550597444176674,
      "epoch": 2.26,
      "learning_rate": 3.8221766626418155e-05,
      "loss": 0.0478,
      "step": 2381,
      "task_loss": 0.06875459849834442
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.763085013986345,
      "compression_loss": 0.0,
      "distillation_loss": 0.057246215641498566,
      "epoch": 2.26,
      "learning_rate": 3.821272229281139e-05,
      "loss": 0.0576,
      "step": 2382,
      "task_loss": 0.06067638099193573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7631033371094456,
      "compression_loss": 0.0,
      "distillation_loss": 0.17733870446681976,
      "epoch": 2.26,
      "learning_rate": 3.820367555904999e-05,
      "loss": 0.1736,
      "step": 2383,
      "task_loss": 0.13981691002845764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7631216541682988,
      "compression_loss": 0.0,
      "distillation_loss": 0.14336426556110382,
      "epoch": 2.26,
      "learning_rate": 3.819462642677733e-05,
      "loss": 0.1475,
      "step": 2384,
      "task_loss": 0.184513121843338
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7631399651639086,
      "compression_loss": 0.0,
      "distillation_loss": 0.1907598078250885,
      "epoch": 2.26,
      "learning_rate": 3.818557489763724e-05,
      "loss": 0.1909,
      "step": 2385,
      "task_loss": 0.19197434186935425
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7631582700972785,
      "compression_loss": 0.0,
      "distillation_loss": 0.09918617457151413,
      "epoch": 2.27,
      "learning_rate": 3.817652097327396e-05,
      "loss": 0.0916,
      "step": 2386,
      "task_loss": 0.023263249546289444
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7631765689694123,
      "compression_loss": 0.0,
      "distillation_loss": 0.0876680240035057,
      "epoch": 2.27,
      "learning_rate": 3.81674646553322e-05,
      "loss": 0.0851,
      "step": 2387,
      "task_loss": 0.06229857727885246
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7631948617813136,
      "compression_loss": 0.0,
      "distillation_loss": 0.09220877289772034,
      "epoch": 2.27,
      "learning_rate": 3.815840594545706e-05,
      "loss": 0.0941,
      "step": 2388,
      "task_loss": 0.11108911037445068
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.763213148533986,
      "compression_loss": 0.0,
      "distillation_loss": 0.10177421569824219,
      "epoch": 2.27,
      "learning_rate": 3.814934484529411e-05,
      "loss": 0.1073,
      "step": 2389,
      "task_loss": 0.157059907913208
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7632314292284335,
      "compression_loss": 0.0,
      "distillation_loss": 0.05992235243320465,
      "epoch": 2.27,
      "learning_rate": 3.8140281356489346e-05,
      "loss": 0.0608,
      "step": 2390,
      "task_loss": 0.06858600676059723
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7632497038656594,
      "compression_loss": 0.0,
      "distillation_loss": 0.0574650838971138,
      "epoch": 2.27,
      "learning_rate": 3.8131215480689184e-05,
      "loss": 0.0586,
      "step": 2391,
      "task_loss": 0.06853343546390533
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7632679724466676,
      "compression_loss": 0.0,
      "distillation_loss": 0.21620208024978638,
      "epoch": 2.27,
      "learning_rate": 3.812214721954049e-05,
      "loss": 0.2103,
      "step": 2392,
      "task_loss": 0.15719348192214966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7632862349724616,
      "compression_loss": 0.0,
      "distillation_loss": 0.17695803940296173,
      "epoch": 2.27,
      "learning_rate": 3.811307657469055e-05,
      "loss": 0.1758,
      "step": 2393,
      "task_loss": 0.16526271402835846
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7633044914440454,
      "compression_loss": 0.0,
      "distillation_loss": 0.05041082203388214,
      "epoch": 2.27,
      "learning_rate": 3.8104003547787105e-05,
      "loss": 0.0526,
      "step": 2394,
      "task_loss": 0.07256819307804108
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7633227418624224,
      "compression_loss": 0.0,
      "distillation_loss": 0.03695048391819,
      "epoch": 2.27,
      "learning_rate": 3.809492814047831e-05,
      "loss": 0.0389,
      "step": 2395,
      "task_loss": 0.05642509087920189
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7633409862285965,
      "compression_loss": 0.0,
      "distillation_loss": 0.1277005672454834,
      "epoch": 2.28,
      "learning_rate": 3.8085850354412745e-05,
      "loss": 0.1204,
      "step": 2396,
      "task_loss": 0.05449753627181053
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7633592245435712,
      "compression_loss": 0.0,
      "distillation_loss": 0.1139630377292633,
      "epoch": 2.28,
      "learning_rate": 3.807677019123944e-05,
      "loss": 0.1069,
      "step": 2397,
      "task_loss": 0.042952507734298706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7633774568083502,
      "compression_loss": 0.0,
      "distillation_loss": 0.24451753497123718,
      "epoch": 2.28,
      "learning_rate": 3.806768765260785e-05,
      "loss": 0.2416,
      "step": 2398,
      "task_loss": 0.2154046595096588
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7633956830239372,
      "compression_loss": 0.0,
      "distillation_loss": 0.20844148099422455,
      "epoch": 2.28,
      "learning_rate": 3.805860274016787e-05,
      "loss": 0.2083,
      "step": 2399,
      "task_loss": 0.20691928267478943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.763413903191336,
      "compression_loss": 0.0,
      "distillation_loss": 0.3789367079734802,
      "epoch": 2.28,
      "learning_rate": 3.8049515455569816e-05,
      "loss": 0.3638,
      "step": 2400,
      "task_loss": 0.22742155194282532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7634321173115501,
      "compression_loss": 0.0,
      "distillation_loss": 0.18300700187683105,
      "epoch": 2.28,
      "learning_rate": 3.804042580046442e-05,
      "loss": 0.1727,
      "step": 2401,
      "task_loss": 0.08020967990159988
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7634503253855833,
      "compression_loss": 0.0,
      "distillation_loss": 0.14992451667785645,
      "epoch": 2.28,
      "learning_rate": 3.803133377650288e-05,
      "loss": 0.1407,
      "step": 2402,
      "task_loss": 0.05811336264014244
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7634685274144393,
      "compression_loss": 0.0,
      "distillation_loss": 0.20229187607765198,
      "epoch": 2.28,
      "learning_rate": 3.80222393853368e-05,
      "loss": 0.1983,
      "step": 2403,
      "task_loss": 0.16219428181648254
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7634867233991217,
      "compression_loss": 0.0,
      "distillation_loss": 0.12786564230918884,
      "epoch": 2.28,
      "learning_rate": 3.801314262861822e-05,
      "loss": 0.1255,
      "step": 2404,
      "task_loss": 0.10410147160291672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7635049133406342,
      "compression_loss": 0.0,
      "distillation_loss": 0.23157238960266113,
      "epoch": 2.28,
      "learning_rate": 3.800404350799961e-05,
      "loss": 0.2361,
      "step": 2405,
      "task_loss": 0.27698665857315063
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7635230972399806,
      "compression_loss": 0.0,
      "distillation_loss": 0.06388011574745178,
      "epoch": 2.28,
      "learning_rate": 3.799494202513386e-05,
      "loss": 0.0808,
      "step": 2406,
      "task_loss": 0.23331348598003387
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7635412750981644,
      "compression_loss": 0.0,
      "distillation_loss": 0.11542148143053055,
      "epoch": 2.29,
      "learning_rate": 3.798583818167432e-05,
      "loss": 0.1163,
      "step": 2407,
      "task_loss": 0.12456656247377396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7635594469161894,
      "compression_loss": 0.0,
      "distillation_loss": 0.08499371260404587,
      "epoch": 2.29,
      "learning_rate": 3.797673197927473e-05,
      "loss": 0.0945,
      "step": 2408,
      "task_loss": 0.17979812622070312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7635776126950591,
      "compression_loss": 0.0,
      "distillation_loss": 0.1530931293964386,
      "epoch": 2.29,
      "learning_rate": 3.796762341958927e-05,
      "loss": 0.1582,
      "step": 2409,
      "task_loss": 0.20382657647132874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7635957724357775,
      "compression_loss": 0.0,
      "distillation_loss": 0.07288970053195953,
      "epoch": 2.29,
      "learning_rate": 3.795851250427257e-05,
      "loss": 0.074,
      "step": 2410,
      "task_loss": 0.08438847959041595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.763613926139348,
      "compression_loss": 0.0,
      "distillation_loss": 0.23029862344264984,
      "epoch": 2.29,
      "learning_rate": 3.794939923497967e-05,
      "loss": 0.2204,
      "step": 2411,
      "task_loss": 0.13161104917526245
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7636320738067744,
      "compression_loss": 0.0,
      "distillation_loss": 0.2242552489042282,
      "epoch": 2.29,
      "learning_rate": 3.794028361336603e-05,
      "loss": 0.2215,
      "step": 2412,
      "task_loss": 0.1969766765832901
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7636502154390605,
      "compression_loss": 0.0,
      "distillation_loss": 0.13051095604896545,
      "epoch": 2.29,
      "learning_rate": 3.793116564108754e-05,
      "loss": 0.122,
      "step": 2413,
      "task_loss": 0.04499632865190506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7636683510372098,
      "compression_loss": 0.0,
      "distillation_loss": 0.03876876085996628,
      "epoch": 2.29,
      "learning_rate": 3.7922045319800545e-05,
      "loss": 0.0358,
      "step": 2414,
      "task_loss": 0.008744785562157631
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.763686480602226,
      "compression_loss": 0.0,
      "distillation_loss": 0.05910433828830719,
      "epoch": 2.29,
      "learning_rate": 3.7912922651161783e-05,
      "loss": 0.0699,
      "step": 2415,
      "task_loss": 0.16721457242965698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7637046041351129,
      "compression_loss": 0.0,
      "distillation_loss": 0.037029679864645004,
      "epoch": 2.29,
      "learning_rate": 3.790379763682844e-05,
      "loss": 0.0346,
      "step": 2416,
      "task_loss": 0.01284867525100708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7637227216368739,
      "compression_loss": 0.0,
      "distillation_loss": 0.25016137957572937,
      "epoch": 2.3,
      "learning_rate": 3.7894670278458096e-05,
      "loss": 0.2403,
      "step": 2417,
      "task_loss": 0.15121634304523468
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7637408331085132,
      "compression_loss": 0.0,
      "distillation_loss": 0.033061422407627106,
      "epoch": 2.3,
      "learning_rate": 3.7885540577708804e-05,
      "loss": 0.036,
      "step": 2418,
      "task_loss": 0.06259030848741531
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7637589385510339,
      "compression_loss": 0.0,
      "distillation_loss": 0.027268044650554657,
      "epoch": 2.3,
      "learning_rate": 3.7876408536239006e-05,
      "loss": 0.0253,
      "step": 2419,
      "task_loss": 0.007266120985150337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.76377703796544,
      "compression_loss": 0.0,
      "distillation_loss": 0.14146488904953003,
      "epoch": 2.3,
      "learning_rate": 3.7867274155707585e-05,
      "loss": 0.1379,
      "step": 2420,
      "task_loss": 0.10578147321939468
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7637951313527352,
      "compression_loss": 0.0,
      "distillation_loss": 0.25725674629211426,
      "epoch": 2.3,
      "learning_rate": 3.7858137437773845e-05,
      "loss": 0.2562,
      "step": 2421,
      "task_loss": 0.2463308572769165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.763813218713923,
      "compression_loss": 0.0,
      "distillation_loss": 0.15272045135498047,
      "epoch": 2.3,
      "learning_rate": 3.784899838409751e-05,
      "loss": 0.1586,
      "step": 2422,
      "task_loss": 0.2110539674758911
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7638313000500074,
      "compression_loss": 0.0,
      "distillation_loss": 0.040354013442993164,
      "epoch": 2.3,
      "learning_rate": 3.783985699633874e-05,
      "loss": 0.0399,
      "step": 2423,
      "task_loss": 0.03630523383617401
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7638493753619917,
      "compression_loss": 0.0,
      "distillation_loss": 0.1408417522907257,
      "epoch": 2.3,
      "learning_rate": 3.783071327615811e-05,
      "loss": 0.1307,
      "step": 2424,
      "task_loss": 0.03991064056754112
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7638674446508799,
      "compression_loss": 0.0,
      "distillation_loss": 0.0335991308093071,
      "epoch": 2.3,
      "learning_rate": 3.7821567225216615e-05,
      "loss": 0.0307,
      "step": 2425,
      "task_loss": 0.004980664700269699
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7638855079176754,
      "compression_loss": 0.0,
      "distillation_loss": 0.13876627385616302,
      "epoch": 2.3,
      "learning_rate": 3.781241884517569e-05,
      "loss": 0.1474,
      "step": 2426,
      "task_loss": 0.2252703756093979
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7639035651633821,
      "compression_loss": 0.0,
      "distillation_loss": 0.18450605869293213,
      "epoch": 2.3,
      "learning_rate": 3.780326813769717e-05,
      "loss": 0.1689,
      "step": 2427,
      "task_loss": 0.028058160096406937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7639216163890036,
      "compression_loss": 0.0,
      "distillation_loss": 0.42347198724746704,
      "epoch": 2.31,
      "learning_rate": 3.779411510444334e-05,
      "loss": 0.4095,
      "step": 2428,
      "task_loss": 0.2832689583301544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7639396615955436,
      "compression_loss": 0.0,
      "distillation_loss": 0.09357698261737823,
      "epoch": 2.31,
      "learning_rate": 3.778495974707688e-05,
      "loss": 0.0918,
      "step": 2429,
      "task_loss": 0.07574363052845001
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7639577007840057,
      "compression_loss": 0.0,
      "distillation_loss": 0.15994183719158173,
      "epoch": 2.31,
      "learning_rate": 3.7775802067260905e-05,
      "loss": 0.1569,
      "step": 2430,
      "task_loss": 0.1296941488981247
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7639757339553938,
      "compression_loss": 0.0,
      "distillation_loss": 0.078113853931427,
      "epoch": 2.31,
      "learning_rate": 3.776664206665896e-05,
      "loss": 0.0896,
      "step": 2431,
      "task_loss": 0.19323307275772095
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7639937611107113,
      "compression_loss": 0.0,
      "distillation_loss": 0.028910305351018906,
      "epoch": 2.31,
      "learning_rate": 3.7757479746935e-05,
      "loss": 0.0324,
      "step": 2432,
      "task_loss": 0.06403736025094986
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.764011782250962,
      "compression_loss": 0.0,
      "distillation_loss": 0.03515395522117615,
      "epoch": 2.31,
      "learning_rate": 3.77483151097534e-05,
      "loss": 0.0325,
      "step": 2433,
      "task_loss": 0.008116895332932472
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7640297973771497,
      "compression_loss": 0.0,
      "distillation_loss": 0.07505205273628235,
      "epoch": 2.31,
      "learning_rate": 3.773914815677897e-05,
      "loss": 0.0719,
      "step": 2434,
      "task_loss": 0.04383193701505661
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7640478064902779,
      "compression_loss": 0.0,
      "distillation_loss": 0.204584002494812,
      "epoch": 2.31,
      "learning_rate": 3.7729978889676915e-05,
      "loss": 0.1907,
      "step": 2435,
      "task_loss": 0.06581425666809082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7640658095913504,
      "compression_loss": 0.0,
      "distillation_loss": 0.1545599400997162,
      "epoch": 2.31,
      "learning_rate": 3.7720807310112896e-05,
      "loss": 0.1583,
      "step": 2436,
      "task_loss": 0.19169500470161438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7640838066813709,
      "compression_loss": 0.0,
      "distillation_loss": 0.10692571848630905,
      "epoch": 2.31,
      "learning_rate": 3.7711633419752954e-05,
      "loss": 0.1141,
      "step": 2437,
      "task_loss": 0.17911353707313538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7641017977613429,
      "compression_loss": 0.0,
      "distillation_loss": 0.2065647542476654,
      "epoch": 2.32,
      "learning_rate": 3.7702457220263595e-05,
      "loss": 0.206,
      "step": 2438,
      "task_loss": 0.20088602602481842
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7641197828322703,
      "compression_loss": 0.0,
      "distillation_loss": 0.1430378258228302,
      "epoch": 2.32,
      "learning_rate": 3.76932787133117e-05,
      "loss": 0.1612,
      "step": 2439,
      "task_loss": 0.3245554566383362
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7641377618951566,
      "compression_loss": 0.0,
      "distillation_loss": 0.12717294692993164,
      "epoch": 2.32,
      "learning_rate": 3.768409790056459e-05,
      "loss": 0.1219,
      "step": 2440,
      "task_loss": 0.07445275783538818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7641557349510057,
      "compression_loss": 0.0,
      "distillation_loss": 0.06758947670459747,
      "epoch": 2.32,
      "learning_rate": 3.7674914783690006e-05,
      "loss": 0.0633,
      "step": 2441,
      "task_loss": 0.024319326505064964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7641737020008211,
      "compression_loss": 0.0,
      "distillation_loss": 0.0334925502538681,
      "epoch": 2.32,
      "learning_rate": 3.7665729364356115e-05,
      "loss": 0.0313,
      "step": 2442,
      "task_loss": 0.011463357135653496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7641916630456065,
      "compression_loss": 0.0,
      "distillation_loss": 0.112037792801857,
      "epoch": 2.32,
      "learning_rate": 3.7656541644231494e-05,
      "loss": 0.1136,
      "step": 2443,
      "task_loss": 0.12805123627185822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7642096180863657,
      "compression_loss": 0.0,
      "distillation_loss": 0.08280123025178909,
      "epoch": 2.32,
      "learning_rate": 3.764735162498512e-05,
      "loss": 0.0867,
      "step": 2444,
      "task_loss": 0.1218644231557846
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7642275671241022,
      "compression_loss": 0.0,
      "distillation_loss": 0.16756752133369446,
      "epoch": 2.32,
      "learning_rate": 3.763815930828641e-05,
      "loss": 0.1726,
      "step": 2445,
      "task_loss": 0.21819201111793518
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7642455101598198,
      "compression_loss": 0.0,
      "distillation_loss": 0.18085908889770508,
      "epoch": 2.32,
      "learning_rate": 3.76289646958052e-05,
      "loss": 0.1723,
      "step": 2446,
      "task_loss": 0.09551871567964554
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7642634471945222,
      "compression_loss": 0.0,
      "distillation_loss": 0.11347385495901108,
      "epoch": 2.32,
      "learning_rate": 3.761976778921173e-05,
      "loss": 0.1107,
      "step": 2447,
      "task_loss": 0.08610384166240692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.764281378229213,
      "compression_loss": 0.0,
      "distillation_loss": 0.029696688055992126,
      "epoch": 2.32,
      "learning_rate": 3.761056859017667e-05,
      "loss": 0.0276,
      "step": 2448,
      "task_loss": 0.009167637676000595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7642993032648961,
      "compression_loss": 0.0,
      "distillation_loss": 0.1404319703578949,
      "epoch": 2.33,
      "learning_rate": 3.7601367100371085e-05,
      "loss": 0.1482,
      "step": 2449,
      "task_loss": 0.21834945678710938
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7643172223025749,
      "compression_loss": 0.0,
      "distillation_loss": 0.15198810398578644,
      "epoch": 2.33,
      "learning_rate": 3.759216332146649e-05,
      "loss": 0.1448,
      "step": 2450,
      "task_loss": 0.07967476546764374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7643351353432531,
      "compression_loss": 0.0,
      "distillation_loss": 0.0767270177602768,
      "epoch": 2.33,
      "learning_rate": 3.7582957255134765e-05,
      "loss": 0.0727,
      "step": 2451,
      "task_loss": 0.03611510992050171
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7643530423879347,
      "compression_loss": 0.0,
      "distillation_loss": 0.06880663335323334,
      "epoch": 2.33,
      "learning_rate": 3.7573748903048266e-05,
      "loss": 0.0728,
      "step": 2452,
      "task_loss": 0.10889902710914612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7643709434376229,
      "compression_loss": 0.0,
      "distillation_loss": 0.11665643751621246,
      "epoch": 2.33,
      "learning_rate": 3.756453826687972e-05,
      "loss": 0.1273,
      "step": 2453,
      "task_loss": 0.2232402116060257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7643888384933218,
      "compression_loss": 0.0,
      "distillation_loss": 0.11299058794975281,
      "epoch": 2.33,
      "learning_rate": 3.755532534830229e-05,
      "loss": 0.1148,
      "step": 2454,
      "task_loss": 0.13139456510543823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7644067275560349,
      "compression_loss": 0.0,
      "distillation_loss": 0.14656969904899597,
      "epoch": 2.33,
      "learning_rate": 3.7546110148989535e-05,
      "loss": 0.1399,
      "step": 2455,
      "task_loss": 0.07960043847560883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7644246106267659,
      "compression_loss": 0.0,
      "distillation_loss": 0.09302163124084473,
      "epoch": 2.33,
      "learning_rate": 3.7536892670615454e-05,
      "loss": 0.0898,
      "step": 2456,
      "task_loss": 0.06115756928920746
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7644424877065185,
      "compression_loss": 0.0,
      "distillation_loss": 0.27814599871635437,
      "epoch": 2.33,
      "learning_rate": 3.752767291485444e-05,
      "loss": 0.29,
      "step": 2457,
      "task_loss": 0.3964260220527649
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7644603587962964,
      "compression_loss": 0.0,
      "distillation_loss": 0.13266442716121674,
      "epoch": 2.33,
      "learning_rate": 3.7518450883381306e-05,
      "loss": 0.1348,
      "step": 2458,
      "task_loss": 0.15446260571479797
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7644782238971031,
      "compression_loss": 0.0,
      "distillation_loss": 0.24767599999904633,
      "epoch": 2.34,
      "learning_rate": 3.750922657787128e-05,
      "loss": 0.2381,
      "step": 2459,
      "task_loss": 0.15177518129348755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7644960830099425,
      "compression_loss": 0.0,
      "distillation_loss": 0.044840142130851746,
      "epoch": 2.34,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.051,
      "step": 2460,
      "task_loss": 0.10653319954872131
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7645139361358183,
      "compression_loss": 0.0,
      "distillation_loss": 0.039941057562828064,
      "epoch": 2.34,
      "learning_rate": 3.7490771151443525e-05,
      "loss": 0.0537,
      "step": 2461,
      "task_loss": 0.17739588022232056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.764531783275734,
      "compression_loss": 0.0,
      "distillation_loss": 0.03993482142686844,
      "epoch": 2.34,
      "learning_rate": 3.748154003387831e-05,
      "loss": 0.0367,
      "step": 2462,
      "task_loss": 0.00717165507376194
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7645496244306934,
      "compression_loss": 0.0,
      "distillation_loss": 0.12393603473901749,
      "epoch": 2.34,
      "learning_rate": 3.7472306648981235e-05,
      "loss": 0.1282,
      "step": 2463,
      "task_loss": 0.16707316040992737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7645674596017003,
      "compression_loss": 0.0,
      "distillation_loss": 0.07649366557598114,
      "epoch": 2.34,
      "learning_rate": 3.746307099842959e-05,
      "loss": 0.0711,
      "step": 2464,
      "task_loss": 0.022800806909799576
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7645852887897581,
      "compression_loss": 0.0,
      "distillation_loss": 0.20021630823612213,
      "epoch": 2.34,
      "learning_rate": 3.745383308390108e-05,
      "loss": 0.2007,
      "step": 2465,
      "task_loss": 0.20535027980804443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7646031119958706,
      "compression_loss": 0.0,
      "distillation_loss": 0.2035316526889801,
      "epoch": 2.34,
      "learning_rate": 3.74445929070738e-05,
      "loss": 0.196,
      "step": 2466,
      "task_loss": 0.1283217966556549
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7646209292210415,
      "compression_loss": 0.0,
      "distillation_loss": 0.06502929329872131,
      "epoch": 2.34,
      "learning_rate": 3.74353504696263e-05,
      "loss": 0.0669,
      "step": 2467,
      "task_loss": 0.08407002687454224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7646387404662746,
      "compression_loss": 0.0,
      "distillation_loss": 0.2670043408870697,
      "epoch": 2.34,
      "learning_rate": 3.742610577323749e-05,
      "loss": 0.275,
      "step": 2468,
      "task_loss": 0.34709686040878296
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7646565457325735,
      "compression_loss": 0.0,
      "distillation_loss": 0.06426650285720825,
      "epoch": 2.34,
      "learning_rate": 3.7416858819586724e-05,
      "loss": 0.0684,
      "step": 2469,
      "task_loss": 0.10525249689817429
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7646743450209417,
      "compression_loss": 0.0,
      "distillation_loss": 0.15390363335609436,
      "epoch": 2.35,
      "learning_rate": 3.740760961035375e-05,
      "loss": 0.1644,
      "step": 2470,
      "task_loss": 0.258728951215744
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7646921383323831,
      "compression_loss": 0.0,
      "distillation_loss": 0.040257424116134644,
      "epoch": 2.35,
      "learning_rate": 3.739835814721874e-05,
      "loss": 0.037,
      "step": 2471,
      "task_loss": 0.0077125392854213715
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7647099256679013,
      "compression_loss": 0.0,
      "distillation_loss": 0.12947189807891846,
      "epoch": 2.35,
      "learning_rate": 3.738910443186226e-05,
      "loss": 0.1284,
      "step": 2472,
      "task_loss": 0.1188727617263794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7647277070285,
      "compression_loss": 0.0,
      "distillation_loss": 0.16135096549987793,
      "epoch": 2.35,
      "learning_rate": 3.737984846596528e-05,
      "loss": 0.1543,
      "step": 2473,
      "task_loss": 0.09042131900787354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7647454824151829,
      "compression_loss": 0.0,
      "distillation_loss": 0.02904283069074154,
      "epoch": 2.35,
      "learning_rate": 3.737059025120922e-05,
      "loss": 0.0267,
      "step": 2474,
      "task_loss": 0.005899908021092415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7647632518289538,
      "compression_loss": 0.0,
      "distillation_loss": 0.11024047434329987,
      "epoch": 2.35,
      "learning_rate": 3.7361329789275855e-05,
      "loss": 0.1193,
      "step": 2475,
      "task_loss": 0.20116351544857025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7647810152708161,
      "compression_loss": 0.0,
      "distillation_loss": 0.04543491452932358,
      "epoch": 2.35,
      "learning_rate": 3.7352067081847405e-05,
      "loss": 0.042,
      "step": 2476,
      "task_loss": 0.010787051171064377
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7647987727417737,
      "compression_loss": 0.0,
      "distillation_loss": 0.09550483524799347,
      "epoch": 2.35,
      "learning_rate": 3.734280213060649e-05,
      "loss": 0.1008,
      "step": 2477,
      "task_loss": 0.1483430564403534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7648165242428301,
      "compression_loss": 0.0,
      "distillation_loss": 0.08557014912366867,
      "epoch": 2.35,
      "learning_rate": 3.7333534937236105e-05,
      "loss": 0.0859,
      "step": 2478,
      "task_loss": 0.08859314769506454
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7648342697749891,
      "compression_loss": 0.0,
      "distillation_loss": 0.12534424662590027,
      "epoch": 2.35,
      "learning_rate": 3.7324265503419716e-05,
      "loss": 0.123,
      "step": 2479,
      "task_loss": 0.10213040560483932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7648520093392545,
      "compression_loss": 0.0,
      "distillation_loss": 0.09746745228767395,
      "epoch": 2.36,
      "learning_rate": 3.731499383084114e-05,
      "loss": 0.1161,
      "step": 2480,
      "task_loss": 0.2837420403957367
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7648697429366297,
      "compression_loss": 0.0,
      "distillation_loss": 0.09911562502384186,
      "epoch": 2.36,
      "learning_rate": 3.730571992118462e-05,
      "loss": 0.1044,
      "step": 2481,
      "task_loss": 0.1517096608877182
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7648874705681187,
      "compression_loss": 0.0,
      "distillation_loss": 0.04106239229440689,
      "epoch": 2.36,
      "learning_rate": 3.7296443776134814e-05,
      "loss": 0.046,
      "step": 2482,
      "task_loss": 0.0909072607755661
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7649051922347249,
      "compression_loss": 0.0,
      "distillation_loss": 0.16157306730747223,
      "epoch": 2.36,
      "learning_rate": 3.7287165397376775e-05,
      "loss": 0.17,
      "step": 2483,
      "task_loss": 0.2461170256137848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7649229079374521,
      "compression_loss": 0.0,
      "distillation_loss": 0.034094762057065964,
      "epoch": 2.36,
      "learning_rate": 3.727788478659597e-05,
      "loss": 0.0376,
      "step": 2484,
      "task_loss": 0.06928237527608871
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7649406176773041,
      "compression_loss": 0.0,
      "distillation_loss": 0.06985623389482498,
      "epoch": 2.36,
      "learning_rate": 3.726860194547826e-05,
      "loss": 0.0685,
      "step": 2485,
      "task_loss": 0.05666497349739075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7649583214552844,
      "compression_loss": 0.0,
      "distillation_loss": 0.07079368084669113,
      "epoch": 2.36,
      "learning_rate": 3.725931687570992e-05,
      "loss": 0.0717,
      "step": 2486,
      "task_loss": 0.07995907217264175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7649760192723968,
      "compression_loss": 0.0,
      "distillation_loss": 0.056753143668174744,
      "epoch": 2.36,
      "learning_rate": 3.7250029578977625e-05,
      "loss": 0.056,
      "step": 2487,
      "task_loss": 0.04943781718611717
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7649937111296448,
      "compression_loss": 0.0,
      "distillation_loss": 0.08018404990434647,
      "epoch": 2.36,
      "learning_rate": 3.724074005696847e-05,
      "loss": 0.0841,
      "step": 2488,
      "task_loss": 0.11959824711084366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7650113970280323,
      "compression_loss": 0.0,
      "distillation_loss": 0.17123107612133026,
      "epoch": 2.36,
      "learning_rate": 3.723144831136992e-05,
      "loss": 0.1641,
      "step": 2489,
      "task_loss": 0.10034439712762833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7650290769685629,
      "compression_loss": 0.0,
      "distillation_loss": 0.06683574616909027,
      "epoch": 2.36,
      "learning_rate": 3.722215434386988e-05,
      "loss": 0.0654,
      "step": 2490,
      "task_loss": 0.05212263762950897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7650467509522403,
      "compression_loss": 0.0,
      "distillation_loss": 0.015314219519495964,
      "epoch": 2.37,
      "learning_rate": 3.721285815615665e-05,
      "loss": 0.0142,
      "step": 2491,
      "task_loss": 0.003950970247387886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.765064418980068,
      "compression_loss": 0.0,
      "distillation_loss": 0.04659897834062576,
      "epoch": 2.37,
      "learning_rate": 3.7203559749918904e-05,
      "loss": 0.0624,
      "step": 2492,
      "task_loss": 0.20480214059352875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7650820810530501,
      "compression_loss": 0.0,
      "distillation_loss": 0.12938767671585083,
      "epoch": 2.37,
      "learning_rate": 3.7194259126845764e-05,
      "loss": 0.1231,
      "step": 2493,
      "task_loss": 0.0663764700293541
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7650997371721899,
      "compression_loss": 0.0,
      "distillation_loss": 0.07641440629959106,
      "epoch": 2.37,
      "learning_rate": 3.7184956288626724e-05,
      "loss": 0.0879,
      "step": 2494,
      "task_loss": 0.1917191445827484
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7651173873384912,
      "compression_loss": 0.0,
      "distillation_loss": 0.08198924362659454,
      "epoch": 2.37,
      "learning_rate": 3.71756512369517e-05,
      "loss": 0.0815,
      "step": 2495,
      "task_loss": 0.077280193567276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7651350315529577,
      "compression_loss": 0.0,
      "distillation_loss": 0.15662863850593567,
      "epoch": 2.37,
      "learning_rate": 3.716634397351097e-05,
      "loss": 0.1554,
      "step": 2496,
      "task_loss": 0.14391350746154785
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.765152669816593,
      "compression_loss": 0.0,
      "distillation_loss": 0.0537070631980896,
      "epoch": 2.37,
      "learning_rate": 3.715703449999528e-05,
      "loss": 0.0495,
      "step": 2497,
      "task_loss": 0.011994145810604095
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7651703021304009,
      "compression_loss": 0.0,
      "distillation_loss": 0.20824488997459412,
      "epoch": 2.37,
      "learning_rate": 3.7147722818095724e-05,
      "loss": 0.2049,
      "step": 2498,
      "task_loss": 0.17516371607780457
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7651879284953851,
      "compression_loss": 0.0,
      "distillation_loss": 0.0815531462430954,
      "epoch": 2.37,
      "learning_rate": 3.713840892950381e-05,
      "loss": 0.089,
      "step": 2499,
      "task_loss": 0.15595076978206635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7652055489125491,
      "compression_loss": 0.0,
      "distillation_loss": 0.04428771510720253,
      "epoch": 2.37,
      "learning_rate": 3.712909283591145e-05,
      "loss": 0.0555,
      "step": 2500,
      "task_loss": 0.15677018463611603
    },
    {
      "epoch": 2.37,
      "eval_accuracy": 0.8864678899082569,
      "eval_loss": 0.4365707039833069,
      "eval_runtime": 18.2037,
      "eval_samples_per_second": 47.902,
      "eval_steps_per_second": 5.988,
      "step": 2500
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7652231633828968,
      "compression_loss": 0.0,
      "distillation_loss": 0.03173205256462097,
      "epoch": 2.38,
      "learning_rate": 3.7119774539010967e-05,
      "loss": 0.0396,
      "step": 2501,
      "task_loss": 0.10992399603128433
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7652407719074318,
      "compression_loss": 0.0,
      "distillation_loss": 0.03812164068222046,
      "epoch": 2.38,
      "learning_rate": 3.711045404049507e-05,
      "loss": 0.0511,
      "step": 2502,
      "task_loss": 0.16801124811172485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7652583744871576,
      "compression_loss": 0.0,
      "distillation_loss": 0.054551735520362854,
      "epoch": 2.38,
      "learning_rate": 3.710113134205687e-05,
      "loss": 0.0503,
      "step": 2503,
      "task_loss": 0.011674726381897926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7652759711230782,
      "compression_loss": 0.0,
      "distillation_loss": 0.057191766798496246,
      "epoch": 2.38,
      "learning_rate": 3.709180644538988e-05,
      "loss": 0.0556,
      "step": 2504,
      "task_loss": 0.04153808206319809
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7652935618161971,
      "compression_loss": 0.0,
      "distillation_loss": 0.07190201431512833,
      "epoch": 2.38,
      "learning_rate": 3.708247935218802e-05,
      "loss": 0.0705,
      "step": 2505,
      "task_loss": 0.05747076869010925
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.765311146567518,
      "compression_loss": 0.0,
      "distillation_loss": 0.0554814338684082,
      "epoch": 2.38,
      "learning_rate": 3.707315006414559e-05,
      "loss": 0.0563,
      "step": 2506,
      "task_loss": 0.06399840116500854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7653287253780445,
      "compression_loss": 0.0,
      "distillation_loss": 0.14144667983055115,
      "epoch": 2.38,
      "learning_rate": 3.706381858295731e-05,
      "loss": 0.1652,
      "step": 2507,
      "task_loss": 0.37935495376586914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7653462982487805,
      "compression_loss": 0.0,
      "distillation_loss": 0.19897310435771942,
      "epoch": 2.38,
      "learning_rate": 3.705448491031829e-05,
      "loss": 0.2037,
      "step": 2508,
      "task_loss": 0.24580544233322144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7653638651807295,
      "compression_loss": 0.0,
      "distillation_loss": 0.03850369527935982,
      "epoch": 2.38,
      "learning_rate": 3.7045149047924016e-05,
      "loss": 0.0414,
      "step": 2509,
      "task_loss": 0.06746485829353333
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7653814261748952,
      "compression_loss": 0.0,
      "distillation_loss": 0.01753471978008747,
      "epoch": 2.38,
      "learning_rate": 3.703581099747041e-05,
      "loss": 0.0161,
      "step": 2510,
      "task_loss": 0.002795502543449402
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7653989812322815,
      "compression_loss": 0.0,
      "distillation_loss": 0.1509072184562683,
      "epoch": 2.38,
      "learning_rate": 3.702647076065378e-05,
      "loss": 0.1567,
      "step": 2511,
      "task_loss": 0.2087806612253189
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7654165303538917,
      "compression_loss": 0.0,
      "distillation_loss": 0.19133007526397705,
      "epoch": 2.39,
      "learning_rate": 3.701712833917082e-05,
      "loss": 0.1858,
      "step": 2512,
      "task_loss": 0.13650889694690704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7654340735407298,
      "compression_loss": 0.0,
      "distillation_loss": 0.11685886234045029,
      "epoch": 2.39,
      "learning_rate": 3.700778373471861e-05,
      "loss": 0.1192,
      "step": 2513,
      "task_loss": 0.1399974673986435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7654516107937992,
      "compression_loss": 0.0,
      "distillation_loss": 0.04263558238744736,
      "epoch": 2.39,
      "learning_rate": 3.699843694899467e-05,
      "loss": 0.0548,
      "step": 2514,
      "task_loss": 0.16461949050426483
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7654691421141039,
      "compression_loss": 0.0,
      "distillation_loss": 0.13114984333515167,
      "epoch": 2.39,
      "learning_rate": 3.698908798369686e-05,
      "loss": 0.1309,
      "step": 2515,
      "task_loss": 0.12871742248535156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7654866675026474,
      "compression_loss": 0.0,
      "distillation_loss": 0.08641094714403152,
      "epoch": 2.39,
      "learning_rate": 3.697973684052347e-05,
      "loss": 0.0799,
      "step": 2516,
      "task_loss": 0.021780574694275856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7655041869604334,
      "compression_loss": 0.0,
      "distillation_loss": 0.05185171589255333,
      "epoch": 2.39,
      "learning_rate": 3.697038352117321e-05,
      "loss": 0.055,
      "step": 2517,
      "task_loss": 0.08340021222829819
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7655217004884656,
      "compression_loss": 0.0,
      "distillation_loss": 0.0397905595600605,
      "epoch": 2.39,
      "learning_rate": 3.6961028027345114e-05,
      "loss": 0.057,
      "step": 2518,
      "task_loss": 0.21199220418930054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7655392080877477,
      "compression_loss": 0.0,
      "distillation_loss": 0.08754666894674301,
      "epoch": 2.39,
      "learning_rate": 3.695167036073868e-05,
      "loss": 0.0948,
      "step": 2519,
      "task_loss": 0.1603783816099167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7655567097592834,
      "compression_loss": 0.0,
      "distillation_loss": 0.049034878611564636,
      "epoch": 2.39,
      "learning_rate": 3.694231052305376e-05,
      "loss": 0.0531,
      "step": 2520,
      "task_loss": 0.09004518389701843
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7655742055040762,
      "compression_loss": 0.0,
      "distillation_loss": 0.17068904638290405,
      "epoch": 2.39,
      "learning_rate": 3.693294851599063e-05,
      "loss": 0.162,
      "step": 2521,
      "task_loss": 0.08347401022911072
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.76559169532313,
      "compression_loss": 0.0,
      "distillation_loss": 0.2087569534778595,
      "epoch": 2.4,
      "learning_rate": 3.692358434124992e-05,
      "loss": 0.2005,
      "step": 2522,
      "task_loss": 0.12605254352092743
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7656091792174484,
      "compression_loss": 0.0,
      "distillation_loss": 0.18309544026851654,
      "epoch": 2.4,
      "learning_rate": 3.69142180005327e-05,
      "loss": 0.1849,
      "step": 2523,
      "task_loss": 0.20159180462360382
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7656266571880351,
      "compression_loss": 0.0,
      "distillation_loss": 0.09189262241125107,
      "epoch": 2.4,
      "learning_rate": 3.69048494955404e-05,
      "loss": 0.0892,
      "step": 2524,
      "task_loss": 0.06457659602165222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7656441292358938,
      "compression_loss": 0.0,
      "distillation_loss": 0.06329502165317535,
      "epoch": 2.4,
      "learning_rate": 3.689547882797485e-05,
      "loss": 0.0645,
      "step": 2525,
      "task_loss": 0.07518291473388672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.765661595362028,
      "compression_loss": 0.0,
      "distillation_loss": 0.04618072509765625,
      "epoch": 2.4,
      "learning_rate": 3.688610599953828e-05,
      "loss": 0.0424,
      "step": 2526,
      "task_loss": 0.008488666266202927
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7656790555674418,
      "compression_loss": 0.0,
      "distillation_loss": 0.1281333714723587,
      "epoch": 2.4,
      "learning_rate": 3.6876731011933316e-05,
      "loss": 0.1215,
      "step": 2527,
      "task_loss": 0.06217958778142929
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7656965098531384,
      "compression_loss": 0.0,
      "distillation_loss": 0.10788536071777344,
      "epoch": 2.4,
      "learning_rate": 3.686735386686296e-05,
      "loss": 0.1122,
      "step": 2528,
      "task_loss": 0.15063226222991943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7657139582201218,
      "compression_loss": 0.0,
      "distillation_loss": 0.11169138550758362,
      "epoch": 2.4,
      "learning_rate": 3.685797456603062e-05,
      "loss": 0.1213,
      "step": 2529,
      "task_loss": 0.20762555301189423
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7657314006693955,
      "compression_loss": 0.0,
      "distillation_loss": 0.2469841092824936,
      "epoch": 2.4,
      "learning_rate": 3.684859311114009e-05,
      "loss": 0.2493,
      "step": 2530,
      "task_loss": 0.2696739435195923
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7657488372019633,
      "compression_loss": 0.0,
      "distillation_loss": 0.08042797446250916,
      "epoch": 2.4,
      "learning_rate": 3.6839209503895566e-05,
      "loss": 0.0757,
      "step": 2531,
      "task_loss": 0.033532898873090744
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7657662678188288,
      "compression_loss": 0.0,
      "distillation_loss": 0.05556613951921463,
      "epoch": 2.4,
      "learning_rate": 3.6829823746001616e-05,
      "loss": 0.0513,
      "step": 2532,
      "task_loss": 0.013295382261276245
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7657836925209958,
      "compression_loss": 0.0,
      "distillation_loss": 0.13678494095802307,
      "epoch": 2.41,
      "learning_rate": 3.6820435839163205e-05,
      "loss": 0.1351,
      "step": 2533,
      "task_loss": 0.12005842477083206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7658011113094678,
      "compression_loss": 0.0,
      "distillation_loss": 0.07145422697067261,
      "epoch": 2.41,
      "learning_rate": 3.68110457850857e-05,
      "loss": 0.0678,
      "step": 2534,
      "task_loss": 0.03514321148395538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7658185241852488,
      "compression_loss": 0.0,
      "distillation_loss": 0.1274898201227188,
      "epoch": 2.41,
      "learning_rate": 3.680165358547484e-05,
      "loss": 0.1235,
      "step": 2535,
      "task_loss": 0.08788056671619415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7658359311493421,
      "compression_loss": 0.0,
      "distillation_loss": 0.17474448680877686,
      "epoch": 2.41,
      "learning_rate": 3.6792259242036776e-05,
      "loss": 0.1785,
      "step": 2536,
      "task_loss": 0.21268706023693085
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7658533322027515,
      "compression_loss": 0.0,
      "distillation_loss": 0.11320266127586365,
      "epoch": 2.41,
      "learning_rate": 3.678286275647802e-05,
      "loss": 0.1103,
      "step": 2537,
      "task_loss": 0.08437247574329376
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7658707273464809,
      "compression_loss": 0.0,
      "distillation_loss": 0.15101607143878937,
      "epoch": 2.41,
      "learning_rate": 3.677346413050551e-05,
      "loss": 0.15,
      "step": 2538,
      "task_loss": 0.14134559035301208
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7658881165815338,
      "compression_loss": 0.0,
      "distillation_loss": 0.10815154016017914,
      "epoch": 2.41,
      "learning_rate": 3.6764063365826525e-05,
      "loss": 0.1181,
      "step": 2539,
      "task_loss": 0.2076764553785324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7659054999089138,
      "compression_loss": 0.0,
      "distillation_loss": 0.10024487972259521,
      "epoch": 2.41,
      "learning_rate": 3.675466046414878e-05,
      "loss": 0.1026,
      "step": 2540,
      "task_loss": 0.12348996847867966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7659228773296247,
      "compression_loss": 0.0,
      "distillation_loss": 0.033068105578422546,
      "epoch": 2.41,
      "learning_rate": 3.674525542718035e-05,
      "loss": 0.0304,
      "step": 2541,
      "task_loss": 0.006272824481129646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7659402488446703,
      "compression_loss": 0.0,
      "distillation_loss": 0.05263807624578476,
      "epoch": 2.41,
      "learning_rate": 3.6735848256629705e-05,
      "loss": 0.0482,
      "step": 2542,
      "task_loss": 0.008157419040799141
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.765957614455054,
      "compression_loss": 0.0,
      "distillation_loss": 0.06785233318805695,
      "epoch": 2.42,
      "learning_rate": 3.6726438954205714e-05,
      "loss": 0.0744,
      "step": 2543,
      "task_loss": 0.13311488926410675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7659749741617796,
      "compression_loss": 0.0,
      "distillation_loss": 0.034298889338970184,
      "epoch": 2.42,
      "learning_rate": 3.6717027521617595e-05,
      "loss": 0.0315,
      "step": 2544,
      "task_loss": 0.006205489858984947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7659923279658509,
      "compression_loss": 0.0,
      "distillation_loss": 0.19808626174926758,
      "epoch": 2.42,
      "learning_rate": 3.6707613960575006e-05,
      "loss": 0.1921,
      "step": 2545,
      "task_loss": 0.13829779624938965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7660096758682715,
      "compression_loss": 0.0,
      "distillation_loss": 0.11868824064731598,
      "epoch": 2.42,
      "learning_rate": 3.669819827278795e-05,
      "loss": 0.118,
      "step": 2546,
      "task_loss": 0.11179932951927185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.766027017870045,
      "compression_loss": 0.0,
      "distillation_loss": 0.018186789005994797,
      "epoch": 2.42,
      "learning_rate": 3.668878045996685e-05,
      "loss": 0.0167,
      "step": 2547,
      "task_loss": 0.0034119393676519394
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7660443539721753,
      "compression_loss": 0.0,
      "distillation_loss": 0.05021512135863304,
      "epoch": 2.42,
      "learning_rate": 3.667936052382248e-05,
      "loss": 0.0501,
      "step": 2548,
      "task_loss": 0.04938085749745369
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7660616841756658,
      "compression_loss": 0.0,
      "distillation_loss": 0.1335231363773346,
      "epoch": 2.42,
      "learning_rate": 3.666993846606602e-05,
      "loss": 0.1321,
      "step": 2549,
      "task_loss": 0.11888933926820755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7660790084815203,
      "compression_loss": 0.0,
      "distillation_loss": 0.07150552421808243,
      "epoch": 2.42,
      "learning_rate": 3.666051428840904e-05,
      "loss": 0.0767,
      "step": 2550,
      "task_loss": 0.12331639230251312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7660963268907426,
      "compression_loss": 0.0,
      "distillation_loss": 0.3968278169631958,
      "epoch": 2.42,
      "learning_rate": 3.665108799256348e-05,
      "loss": 0.4067,
      "step": 2551,
      "task_loss": 0.4955511689186096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7661136394043363,
      "compression_loss": 0.0,
      "distillation_loss": 0.2079869657754898,
      "epoch": 2.42,
      "learning_rate": 3.6641659580241665e-05,
      "loss": 0.2161,
      "step": 2552,
      "task_loss": 0.2889349162578583
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.766130946023305,
      "compression_loss": 0.0,
      "distillation_loss": 0.21949072182178497,
      "epoch": 2.42,
      "learning_rate": 3.663222905315633e-05,
      "loss": 0.2231,
      "step": 2553,
      "task_loss": 0.25531524419784546
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7661482467486525,
      "compression_loss": 0.0,
      "distillation_loss": 0.1400509923696518,
      "epoch": 2.43,
      "learning_rate": 3.662279641302056e-05,
      "loss": 0.1469,
      "step": 2554,
      "task_loss": 0.2087712436914444
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7661655415813824,
      "compression_loss": 0.0,
      "distillation_loss": 0.029690932482481003,
      "epoch": 2.43,
      "learning_rate": 3.6613361661547854e-05,
      "loss": 0.0333,
      "step": 2555,
      "task_loss": 0.06537258625030518
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7661828305224985,
      "compression_loss": 0.0,
      "distillation_loss": 0.08491934835910797,
      "epoch": 2.43,
      "learning_rate": 3.660392480045206e-05,
      "loss": 0.0799,
      "step": 2556,
      "task_loss": 0.03468838334083557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7662001135730042,
      "compression_loss": 0.0,
      "distillation_loss": 0.06277995556592941,
      "epoch": 2.43,
      "learning_rate": 3.659448583144745e-05,
      "loss": 0.06,
      "step": 2557,
      "task_loss": 0.03468390554189682
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7662173907339035,
      "compression_loss": 0.0,
      "distillation_loss": 0.14954762160778046,
      "epoch": 2.43,
      "learning_rate": 3.658504475624865e-05,
      "loss": 0.1508,
      "step": 2558,
      "task_loss": 0.16203594207763672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7662346620061999,
      "compression_loss": 0.0,
      "distillation_loss": 0.08595126867294312,
      "epoch": 2.43,
      "learning_rate": 3.657560157657067e-05,
      "loss": 0.0885,
      "step": 2559,
      "task_loss": 0.11119415611028671
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7662519273908973,
      "compression_loss": 0.0,
      "distillation_loss": 0.25560837984085083,
      "epoch": 2.43,
      "learning_rate": 3.656615629412892e-05,
      "loss": 0.2518,
      "step": 2560,
      "task_loss": 0.21740317344665527
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.766269186888999,
      "compression_loss": 0.0,
      "distillation_loss": 0.03523785620927811,
      "epoch": 2.43,
      "learning_rate": 3.655670891063917e-05,
      "loss": 0.0325,
      "step": 2561,
      "task_loss": 0.007899057120084763
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7662864405015091,
      "compression_loss": 0.0,
      "distillation_loss": 0.07272191345691681,
      "epoch": 2.43,
      "learning_rate": 3.6547259427817595e-05,
      "loss": 0.068,
      "step": 2562,
      "task_loss": 0.025067364796996117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7663036882294311,
      "compression_loss": 0.0,
      "distillation_loss": 0.06609214842319489,
      "epoch": 2.43,
      "learning_rate": 3.6537807847380726e-05,
      "loss": 0.0712,
      "step": 2563,
      "task_loss": 0.11691683530807495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7663209300737686,
      "compression_loss": 0.0,
      "distillation_loss": 0.06222629174590111,
      "epoch": 2.43,
      "learning_rate": 3.65283541710455e-05,
      "loss": 0.0747,
      "step": 2564,
      "task_loss": 0.18653015792369843
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7663381660355253,
      "compression_loss": 0.0,
      "distillation_loss": 0.08360789716243744,
      "epoch": 2.44,
      "learning_rate": 3.6518898400529214e-05,
      "loss": 0.0781,
      "step": 2565,
      "task_loss": 0.02866414003074169
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7663553961157049,
      "compression_loss": 0.0,
      "distillation_loss": 0.044771708548069,
      "epoch": 2.44,
      "learning_rate": 3.650944053754956e-05,
      "loss": 0.0451,
      "step": 2566,
      "task_loss": 0.04824347794055939
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7663726203153112,
      "compression_loss": 0.0,
      "distillation_loss": 0.09424047917127609,
      "epoch": 2.44,
      "learning_rate": 3.6499980583824606e-05,
      "loss": 0.0961,
      "step": 2567,
      "task_loss": 0.1124841496348381
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7663898386353478,
      "compression_loss": 0.0,
      "distillation_loss": 0.04939156025648117,
      "epoch": 2.44,
      "learning_rate": 3.64905185410728e-05,
      "loss": 0.0606,
      "step": 2568,
      "task_loss": 0.16101084649562836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7664070510768184,
      "compression_loss": 0.0,
      "distillation_loss": 0.08628442138433456,
      "epoch": 2.44,
      "learning_rate": 3.6481054411012946e-05,
      "loss": 0.087,
      "step": 2569,
      "task_loss": 0.09357213228940964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7664242576407266,
      "compression_loss": 0.0,
      "distillation_loss": 0.06562324613332748,
      "epoch": 2.44,
      "learning_rate": 3.647158819536427e-05,
      "loss": 0.0648,
      "step": 2570,
      "task_loss": 0.057407625019550323
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7664414583280762,
      "compression_loss": 0.0,
      "distillation_loss": 0.18060490489006042,
      "epoch": 2.44,
      "learning_rate": 3.646211989584635e-05,
      "loss": 0.1716,
      "step": 2571,
      "task_loss": 0.09014902263879776
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7664586531398708,
      "compression_loss": 0.0,
      "distillation_loss": 0.072050541639328,
      "epoch": 2.44,
      "learning_rate": 3.645264951417915e-05,
      "loss": 0.0875,
      "step": 2572,
      "task_loss": 0.2262389212846756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7664758420771142,
      "compression_loss": 0.0,
      "distillation_loss": 0.1359626054763794,
      "epoch": 2.44,
      "learning_rate": 3.644317705208301e-05,
      "loss": 0.1305,
      "step": 2573,
      "task_loss": 0.08181658387184143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7664930251408099,
      "compression_loss": 0.0,
      "distillation_loss": 0.049975574016571045,
      "epoch": 2.44,
      "learning_rate": 3.643370251127865e-05,
      "loss": 0.0476,
      "step": 2574,
      "task_loss": 0.025839539244771004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7665102023319617,
      "compression_loss": 0.0,
      "distillation_loss": 0.13199934363365173,
      "epoch": 2.45,
      "learning_rate": 3.6424225893487166e-05,
      "loss": 0.1329,
      "step": 2575,
      "task_loss": 0.1408880352973938
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7665273736515732,
      "compression_loss": 0.0,
      "distillation_loss": 0.036083199083805084,
      "epoch": 2.45,
      "learning_rate": 3.641474720043002e-05,
      "loss": 0.0437,
      "step": 2576,
      "task_loss": 0.11259350180625916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7665445391006481,
      "compression_loss": 0.0,
      "distillation_loss": 0.06714160740375519,
      "epoch": 2.45,
      "learning_rate": 3.6405266433829075e-05,
      "loss": 0.0623,
      "step": 2577,
      "task_loss": 0.0186399407684803
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7665616986801902,
      "compression_loss": 0.0,
      "distillation_loss": 0.1377488076686859,
      "epoch": 2.45,
      "learning_rate": 3.639578359540655e-05,
      "loss": 0.1438,
      "step": 2578,
      "task_loss": 0.19833146035671234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7665788523912032,
      "compression_loss": 0.0,
      "distillation_loss": 0.030588299036026,
      "epoch": 2.45,
      "learning_rate": 3.638629868688506e-05,
      "loss": 0.0324,
      "step": 2579,
      "task_loss": 0.04895820468664169
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7665960002346905,
      "compression_loss": 0.0,
      "distillation_loss": 0.022023940458893776,
      "epoch": 2.45,
      "learning_rate": 3.6376811709987574e-05,
      "loss": 0.0295,
      "step": 2580,
      "task_loss": 0.09721846878528595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.766613142211656,
      "compression_loss": 0.0,
      "distillation_loss": 0.021494712680578232,
      "epoch": 2.45,
      "learning_rate": 3.636732266643745e-05,
      "loss": 0.028,
      "step": 2581,
      "task_loss": 0.0862061157822609
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7666302783231035,
      "compression_loss": 0.0,
      "distillation_loss": 0.019449274986982346,
      "epoch": 2.45,
      "learning_rate": 3.635783155795841e-05,
      "loss": 0.0182,
      "step": 2582,
      "task_loss": 0.006459413096308708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7666474085700363,
      "compression_loss": 0.0,
      "distillation_loss": 0.04088941961526871,
      "epoch": 2.45,
      "learning_rate": 3.634833838627458e-05,
      "loss": 0.0504,
      "step": 2583,
      "task_loss": 0.13594874739646912
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7666645329534585,
      "compression_loss": 0.0,
      "distillation_loss": 0.23599717020988464,
      "epoch": 2.45,
      "learning_rate": 3.6338843153110424e-05,
      "loss": 0.2423,
      "step": 2584,
      "task_loss": 0.29853349924087524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7666816514743735,
      "compression_loss": 0.0,
      "distillation_loss": 0.1065201386809349,
      "epoch": 2.45,
      "learning_rate": 3.63293458601908e-05,
      "loss": 0.1064,
      "step": 2585,
      "task_loss": 0.10565009713172913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7666987641337851,
      "compression_loss": 0.0,
      "distillation_loss": 0.08189831674098969,
      "epoch": 2.46,
      "learning_rate": 3.631984650924094e-05,
      "loss": 0.0766,
      "step": 2586,
      "task_loss": 0.029174111783504486
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.766715870932697,
      "compression_loss": 0.0,
      "distillation_loss": 0.07314605265855789,
      "epoch": 2.46,
      "learning_rate": 3.631034510198643e-05,
      "loss": 0.067,
      "step": 2587,
      "task_loss": 0.011908493936061859
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7667329718721128,
      "compression_loss": 0.0,
      "distillation_loss": 0.20526961982250214,
      "epoch": 2.46,
      "learning_rate": 3.630084164015328e-05,
      "loss": 0.2009,
      "step": 2588,
      "task_loss": 0.1615629643201828
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7667500669530363,
      "compression_loss": 0.0,
      "distillation_loss": 0.02173478901386261,
      "epoch": 2.46,
      "learning_rate": 3.6291336125467814e-05,
      "loss": 0.0284,
      "step": 2589,
      "task_loss": 0.08851068466901779
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.766767156176471,
      "compression_loss": 0.0,
      "distillation_loss": 0.0929790809750557,
      "epoch": 2.46,
      "learning_rate": 3.628182855965676e-05,
      "loss": 0.0901,
      "step": 2590,
      "task_loss": 0.06388040632009506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7667842395434207,
      "compression_loss": 0.0,
      "distillation_loss": 0.09808763116598129,
      "epoch": 2.46,
      "learning_rate": 3.627231894444721e-05,
      "loss": 0.093,
      "step": 2591,
      "task_loss": 0.04766422510147095
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7668013170548891,
      "compression_loss": 0.0,
      "distillation_loss": 0.08065304160118103,
      "epoch": 2.46,
      "learning_rate": 3.6262807281566634e-05,
      "loss": 0.0768,
      "step": 2592,
      "task_loss": 0.04211016744375229
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7668183887118799,
      "compression_loss": 0.0,
      "distillation_loss": 0.09210735559463501,
      "epoch": 2.46,
      "learning_rate": 3.6253293572742884e-05,
      "loss": 0.0841,
      "step": 2593,
      "task_loss": 0.011922374367713928
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7668354545153966,
      "compression_loss": 0.0,
      "distillation_loss": 0.20635256171226501,
      "epoch": 2.46,
      "learning_rate": 3.624377781970416e-05,
      "loss": 0.1924,
      "step": 2594,
      "task_loss": 0.06673353165388107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7668525144664431,
      "compression_loss": 0.0,
      "distillation_loss": 0.1058523878455162,
      "epoch": 2.46,
      "learning_rate": 3.6234260024179033e-05,
      "loss": 0.1146,
      "step": 2595,
      "task_loss": 0.1931142807006836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.766869568566023,
      "compression_loss": 0.0,
      "distillation_loss": 0.029114918783307076,
      "epoch": 2.47,
      "learning_rate": 3.622474018789648e-05,
      "loss": 0.0281,
      "step": 2596,
      "task_loss": 0.018587203696370125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.76688661681514,
      "compression_loss": 0.0,
      "distillation_loss": 0.025142014026641846,
      "epoch": 2.47,
      "learning_rate": 3.62152183125858e-05,
      "loss": 0.0326,
      "step": 2597,
      "task_loss": 0.09944605827331543
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7669036592147978,
      "compression_loss": 0.0,
      "distillation_loss": 0.19486957788467407,
      "epoch": 2.47,
      "learning_rate": 3.620569439997671e-05,
      "loss": 0.2059,
      "step": 2598,
      "task_loss": 0.30563467741012573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7669206957659999,
      "compression_loss": 0.0,
      "distillation_loss": 0.03863801434636116,
      "epoch": 2.47,
      "learning_rate": 3.6196168451799266e-05,
      "loss": 0.0356,
      "step": 2599,
      "task_loss": 0.008016956970095634
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7669377264697502,
      "compression_loss": 0.0,
      "distillation_loss": 0.13873320817947388,
      "epoch": 2.47,
      "learning_rate": 3.618664046978389e-05,
      "loss": 0.1375,
      "step": 2600,
      "task_loss": 0.12616637349128723
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7669547513270523,
      "compression_loss": 0.0,
      "distillation_loss": 0.022825509309768677,
      "epoch": 2.47,
      "learning_rate": 3.617711045566141e-05,
      "loss": 0.0215,
      "step": 2601,
      "task_loss": 0.00933530181646347
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.76697177033891,
      "compression_loss": 0.0,
      "distillation_loss": 0.06487306207418442,
      "epoch": 2.47,
      "learning_rate": 3.616757841116298e-05,
      "loss": 0.0745,
      "step": 2602,
      "task_loss": 0.16160070896148682
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7669887835063267,
      "compression_loss": 0.0,
      "distillation_loss": 0.0611257366836071,
      "epoch": 2.47,
      "learning_rate": 3.6158044338020155e-05,
      "loss": 0.0651,
      "step": 2603,
      "task_loss": 0.1010468602180481
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7670057908303063,
      "compression_loss": 0.0,
      "distillation_loss": 0.028607673943042755,
      "epoch": 2.47,
      "learning_rate": 3.614850823796483e-05,
      "loss": 0.0267,
      "step": 2604,
      "task_loss": 0.009802697226405144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7670227923118524,
      "compression_loss": 0.0,
      "distillation_loss": 0.19368760287761688,
      "epoch": 2.47,
      "learning_rate": 3.6138970112729296e-05,
      "loss": 0.1928,
      "step": 2605,
      "task_loss": 0.184406578540802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7670397879519689,
      "compression_loss": 0.0,
      "distillation_loss": 0.06413474678993225,
      "epoch": 2.47,
      "learning_rate": 3.612942996404619e-05,
      "loss": 0.0689,
      "step": 2606,
      "task_loss": 0.11217701435089111
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7670567777516591,
      "compression_loss": 0.0,
      "distillation_loss": 0.07466208934783936,
      "epoch": 2.48,
      "learning_rate": 3.611988779364853e-05,
      "loss": 0.0731,
      "step": 2607,
      "task_loss": 0.059125810861587524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.767073761711927,
      "compression_loss": 0.0,
      "distillation_loss": 0.034310828894376755,
      "epoch": 2.48,
      "learning_rate": 3.611034360326971e-05,
      "loss": 0.0405,
      "step": 2608,
      "task_loss": 0.09609294682741165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7670907398337762,
      "compression_loss": 0.0,
      "distillation_loss": 0.07237622141838074,
      "epoch": 2.48,
      "learning_rate": 3.6100797394643455e-05,
      "loss": 0.0682,
      "step": 2609,
      "task_loss": 0.03029775619506836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7671077121182103,
      "compression_loss": 0.0,
      "distillation_loss": 0.03261907026171684,
      "epoch": 2.48,
      "learning_rate": 3.60912491695039e-05,
      "loss": 0.0302,
      "step": 2610,
      "task_loss": 0.008593011647462845
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7671246785662329,
      "compression_loss": 0.0,
      "distillation_loss": 0.044936779886484146,
      "epoch": 2.48,
      "learning_rate": 3.608169892958551e-05,
      "loss": 0.0455,
      "step": 2611,
      "task_loss": 0.050538793206214905
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7671416391788479,
      "compression_loss": 0.0,
      "distillation_loss": 0.04782456159591675,
      "epoch": 2.48,
      "learning_rate": 3.607214667662314e-05,
      "loss": 0.0502,
      "step": 2612,
      "task_loss": 0.07197590172290802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.767158593957059,
      "compression_loss": 0.0,
      "distillation_loss": 0.14677798748016357,
      "epoch": 2.48,
      "learning_rate": 3.606259241235201e-05,
      "loss": 0.1464,
      "step": 2613,
      "task_loss": 0.1426515430212021
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7671755429018696,
      "compression_loss": 0.0,
      "distillation_loss": 0.08823909610509872,
      "epoch": 2.48,
      "learning_rate": 3.605303613850768e-05,
      "loss": 0.0938,
      "step": 2614,
      "task_loss": 0.14365153014659882
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7671924860142837,
      "compression_loss": 0.0,
      "distillation_loss": 0.21100696921348572,
      "epoch": 2.48,
      "learning_rate": 3.604347785682611e-05,
      "loss": 0.2037,
      "step": 2615,
      "task_loss": 0.13835720717906952
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7672094232953047,
      "compression_loss": 0.0,
      "distillation_loss": 0.12257960438728333,
      "epoch": 2.48,
      "learning_rate": 3.60339175690436e-05,
      "loss": 0.115,
      "step": 2616,
      "task_loss": 0.04694103077054024
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7672263547459366,
      "compression_loss": 0.0,
      "distillation_loss": 0.18178528547286987,
      "epoch": 2.49,
      "learning_rate": 3.602435527689683e-05,
      "loss": 0.1834,
      "step": 2617,
      "task_loss": 0.198032945394516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7672432803671828,
      "compression_loss": 0.0,
      "distillation_loss": 0.11362110078334808,
      "epoch": 2.49,
      "learning_rate": 3.6014790982122816e-05,
      "loss": 0.1279,
      "step": 2618,
      "task_loss": 0.25671568512916565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.767260200160047,
      "compression_loss": 0.0,
      "distillation_loss": 0.10963121801614761,
      "epoch": 2.49,
      "learning_rate": 3.6005224686458985e-05,
      "loss": 0.1102,
      "step": 2619,
      "task_loss": 0.11555326730012894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7672771141255331,
      "compression_loss": 0.0,
      "distillation_loss": 0.02279139682650566,
      "epoch": 2.49,
      "learning_rate": 3.599565639164308e-05,
      "loss": 0.0305,
      "step": 2620,
      "task_loss": 0.09947670251131058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7672940222646447,
      "compression_loss": 0.0,
      "distillation_loss": 0.07615819573402405,
      "epoch": 2.49,
      "learning_rate": 3.5986086099413234e-05,
      "loss": 0.0782,
      "step": 2621,
      "task_loss": 0.09702453017234802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7673109245783853,
      "compression_loss": 0.0,
      "distillation_loss": 0.09059733152389526,
      "epoch": 2.49,
      "learning_rate": 3.597651381150795e-05,
      "loss": 0.1036,
      "step": 2622,
      "task_loss": 0.22066594660282135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7673278210677588,
      "compression_loss": 0.0,
      "distillation_loss": 0.17163929343223572,
      "epoch": 2.49,
      "learning_rate": 3.5966939529666056e-05,
      "loss": 0.1662,
      "step": 2623,
      "task_loss": 0.11702217161655426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7673447117337687,
      "compression_loss": 0.0,
      "distillation_loss": 0.25577402114868164,
      "epoch": 2.49,
      "learning_rate": 3.595736325562679e-05,
      "loss": 0.2671,
      "step": 2624,
      "task_loss": 0.36930835247039795
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7673615965774189,
      "compression_loss": 0.0,
      "distillation_loss": 0.0822661817073822,
      "epoch": 2.49,
      "learning_rate": 3.5947784991129716e-05,
      "loss": 0.0869,
      "step": 2625,
      "task_loss": 0.1290304958820343
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7673784755997128,
      "compression_loss": 0.0,
      "distillation_loss": 0.059237804263830185,
      "epoch": 2.49,
      "learning_rate": 3.593820473791476e-05,
      "loss": 0.0793,
      "step": 2626,
      "task_loss": 0.26018190383911133
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7673953488016544,
      "compression_loss": 0.0,
      "distillation_loss": 0.31930679082870483,
      "epoch": 2.49,
      "learning_rate": 3.5928622497722245e-05,
      "loss": 0.3108,
      "step": 2627,
      "task_loss": 0.2345261424779892
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7674122161842472,
      "compression_loss": 0.0,
      "distillation_loss": 0.12323985993862152,
      "epoch": 2.5,
      "learning_rate": 3.591903827229282e-05,
      "loss": 0.1256,
      "step": 2628,
      "task_loss": 0.1471230387687683
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7674290777484948,
      "compression_loss": 0.0,
      "distillation_loss": 0.16051419079303741,
      "epoch": 2.5,
      "learning_rate": 3.590945206336751e-05,
      "loss": 0.1523,
      "step": 2629,
      "task_loss": 0.07876568287611008
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.767445933495401,
      "compression_loss": 0.0,
      "distillation_loss": 0.15569360554218292,
      "epoch": 2.5,
      "learning_rate": 3.5899863872687675e-05,
      "loss": 0.1511,
      "step": 2630,
      "task_loss": 0.10985402017831802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7674627834259696,
      "compression_loss": 0.0,
      "distillation_loss": 0.3402307629585266,
      "epoch": 2.5,
      "learning_rate": 3.58902737019951e-05,
      "loss": 0.3425,
      "step": 2631,
      "task_loss": 0.36277827620506287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.767479627541204,
      "compression_loss": 0.0,
      "distillation_loss": 0.2227693647146225,
      "epoch": 2.5,
      "learning_rate": 3.5880681553031835e-05,
      "loss": 0.2177,
      "step": 2632,
      "task_loss": 0.17254310846328735
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7674964658421082,
      "compression_loss": 0.0,
      "distillation_loss": 0.1516626626253128,
      "epoch": 2.5,
      "learning_rate": 3.5871087427540375e-05,
      "loss": 0.153,
      "step": 2633,
      "task_loss": 0.16509924829006195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7675132983296856,
      "compression_loss": 0.0,
      "distillation_loss": 0.06550684571266174,
      "epoch": 2.5,
      "learning_rate": 3.586149132726353e-05,
      "loss": 0.0695,
      "step": 2634,
      "task_loss": 0.10565401613712311
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7675301250049401,
      "compression_loss": 0.0,
      "distillation_loss": 0.08065078407526016,
      "epoch": 2.5,
      "learning_rate": 3.585189325394447e-05,
      "loss": 0.0776,
      "step": 2635,
      "task_loss": 0.04991026595234871
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7675469458688752,
      "compression_loss": 0.0,
      "distillation_loss": 0.10402727127075195,
      "epoch": 2.5,
      "learning_rate": 3.5842293209326746e-05,
      "loss": 0.1008,
      "step": 2636,
      "task_loss": 0.07204234600067139
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7675637609224947,
      "compression_loss": 0.0,
      "distillation_loss": 0.1009630411863327,
      "epoch": 2.5,
      "learning_rate": 3.583269119515423e-05,
      "loss": 0.1036,
      "step": 2637,
      "task_loss": 0.1269753873348236
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7675805701668021,
      "compression_loss": 0.0,
      "distillation_loss": 0.08721812069416046,
      "epoch": 2.51,
      "learning_rate": 3.58230872131712e-05,
      "loss": 0.0916,
      "step": 2638,
      "task_loss": 0.13126561045646667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7675973736028014,
      "compression_loss": 0.0,
      "distillation_loss": 0.06688186526298523,
      "epoch": 2.51,
      "learning_rate": 3.581348126512225e-05,
      "loss": 0.0659,
      "step": 2639,
      "task_loss": 0.0573241263628006
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7676141712314961,
      "compression_loss": 0.0,
      "distillation_loss": 0.07377175241708755,
      "epoch": 2.51,
      "learning_rate": 3.5803873352752343e-05,
      "loss": 0.0862,
      "step": 2640,
      "task_loss": 0.19805538654327393
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7676309630538898,
      "compression_loss": 0.0,
      "distillation_loss": 0.1930721551179886,
      "epoch": 2.51,
      "learning_rate": 3.5794263477806816e-05,
      "loss": 0.185,
      "step": 2641,
      "task_loss": 0.1118684709072113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7676477490709863,
      "compression_loss": 0.0,
      "distillation_loss": 0.13114435970783234,
      "epoch": 2.51,
      "learning_rate": 3.578465164203134e-05,
      "loss": 0.1396,
      "step": 2642,
      "task_loss": 0.21528376638889313
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7676645292837894,
      "compression_loss": 0.0,
      "distillation_loss": 0.1114916130900383,
      "epoch": 2.51,
      "learning_rate": 3.577503784717195e-05,
      "loss": 0.1063,
      "step": 2643,
      "task_loss": 0.05928418040275574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7676813036933025,
      "compression_loss": 0.0,
      "distillation_loss": 0.09665559232234955,
      "epoch": 2.51,
      "learning_rate": 3.576542209497505e-05,
      "loss": 0.0976,
      "step": 2644,
      "task_loss": 0.10586561262607574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7676980723005294,
      "compression_loss": 0.0,
      "distillation_loss": 0.057753339409828186,
      "epoch": 2.51,
      "learning_rate": 3.575580438718738e-05,
      "loss": 0.0596,
      "step": 2645,
      "task_loss": 0.07584469765424728
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7677148351064739,
      "compression_loss": 0.0,
      "distillation_loss": 0.11892731487751007,
      "epoch": 2.51,
      "learning_rate": 3.574618472555604e-05,
      "loss": 0.1125,
      "step": 2646,
      "task_loss": 0.05511043220758438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7677315921121395,
      "compression_loss": 0.0,
      "distillation_loss": 0.1064673513174057,
      "epoch": 2.51,
      "learning_rate": 3.573656311182848e-05,
      "loss": 0.1096,
      "step": 2647,
      "task_loss": 0.13810373842716217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7677483433185299,
      "compression_loss": 0.0,
      "distillation_loss": 0.17461617290973663,
      "epoch": 2.51,
      "learning_rate": 3.5726939547752536e-05,
      "loss": 0.1789,
      "step": 2648,
      "task_loss": 0.21704277396202087
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.767765088726649,
      "compression_loss": 0.0,
      "distillation_loss": 0.22404174506664276,
      "epoch": 2.52,
      "learning_rate": 3.5717314035076355e-05,
      "loss": 0.216,
      "step": 2649,
      "task_loss": 0.14349466562271118
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7677818283375003,
      "compression_loss": 0.0,
      "distillation_loss": 0.05500160902738571,
      "epoch": 2.52,
      "learning_rate": 3.570768657554847e-05,
      "loss": 0.0503,
      "step": 2650,
      "task_loss": 0.008138328790664673
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7677985621520874,
      "compression_loss": 0.0,
      "distillation_loss": 0.12913605570793152,
      "epoch": 2.52,
      "learning_rate": 3.569805717091775e-05,
      "loss": 0.1416,
      "step": 2651,
      "task_loss": 0.2541487216949463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7678152901714143,
      "compression_loss": 0.0,
      "distillation_loss": 0.10944856703281403,
      "epoch": 2.52,
      "learning_rate": 3.5688425822933414e-05,
      "loss": 0.1132,
      "step": 2652,
      "task_loss": 0.14650918543338776
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7678320123964844,
      "compression_loss": 0.0,
      "distillation_loss": 0.031936414539813995,
      "epoch": 2.52,
      "learning_rate": 3.5678792533345055e-05,
      "loss": 0.0292,
      "step": 2653,
      "task_loss": 0.004621490836143494
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7678487288283014,
      "compression_loss": 0.0,
      "distillation_loss": 0.0944894552230835,
      "epoch": 2.52,
      "learning_rate": 3.5669157303902604e-05,
      "loss": 0.0947,
      "step": 2654,
      "task_loss": 0.09676090627908707
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.767865439467869,
      "compression_loss": 0.0,
      "distillation_loss": 0.03909418731927872,
      "epoch": 2.52,
      "learning_rate": 3.565952013635635e-05,
      "loss": 0.0408,
      "step": 2655,
      "task_loss": 0.05638463795185089
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.767882144316191,
      "compression_loss": 0.0,
      "distillation_loss": 0.06495656073093414,
      "epoch": 2.52,
      "learning_rate": 3.564988103245692e-05,
      "loss": 0.0731,
      "step": 2656,
      "task_loss": 0.14613646268844604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.767898843374271,
      "compression_loss": 0.0,
      "distillation_loss": 0.09904111921787262,
      "epoch": 2.52,
      "learning_rate": 3.564023999395531e-05,
      "loss": 0.1154,
      "step": 2657,
      "task_loss": 0.26308944821357727
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7679155366431126,
      "compression_loss": 0.0,
      "distillation_loss": 0.13033923506736755,
      "epoch": 2.52,
      "learning_rate": 3.563059702260287e-05,
      "loss": 0.1402,
      "step": 2658,
      "task_loss": 0.22941374778747559
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7679322241237196,
      "compression_loss": 0.0,
      "distillation_loss": 0.038218479603528976,
      "epoch": 2.53,
      "learning_rate": 3.562095212015128e-05,
      "loss": 0.0439,
      "step": 2659,
      "task_loss": 0.09546901285648346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7679489058170957,
      "compression_loss": 0.0,
      "distillation_loss": 0.06592346727848053,
      "epoch": 2.53,
      "learning_rate": 3.5611305288352576e-05,
      "loss": 0.0639,
      "step": 2660,
      "task_loss": 0.04572898894548416
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7679655817242446,
      "compression_loss": 0.0,
      "distillation_loss": 0.1503419280052185,
      "epoch": 2.53,
      "learning_rate": 3.560165652895917e-05,
      "loss": 0.1453,
      "step": 2661,
      "task_loss": 0.09945236146450043
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7679822518461699,
      "compression_loss": 0.0,
      "distillation_loss": 0.03605799376964569,
      "epoch": 2.53,
      "learning_rate": 3.5592005843723794e-05,
      "loss": 0.0407,
      "step": 2662,
      "task_loss": 0.08254842460155487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7679989161838752,
      "compression_loss": 0.0,
      "distillation_loss": 0.06412842869758606,
      "epoch": 2.53,
      "learning_rate": 3.558235323439955e-05,
      "loss": 0.0721,
      "step": 2663,
      "task_loss": 0.14363062381744385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7680155747383642,
      "compression_loss": 0.0,
      "distillation_loss": 0.045804090797901154,
      "epoch": 2.53,
      "learning_rate": 3.557269870273987e-05,
      "loss": 0.0485,
      "step": 2664,
      "task_loss": 0.07302069664001465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7680322275106408,
      "compression_loss": 0.0,
      "distillation_loss": 0.1284482777118683,
      "epoch": 2.53,
      "learning_rate": 3.5563042250498556e-05,
      "loss": 0.131,
      "step": 2665,
      "task_loss": 0.1538638472557068
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7680488745017086,
      "compression_loss": 0.0,
      "distillation_loss": 0.037929221987724304,
      "epoch": 2.53,
      "learning_rate": 3.555338387942974e-05,
      "loss": 0.0348,
      "step": 2666,
      "task_loss": 0.0066341981291770935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7680655157125711,
      "compression_loss": 0.0,
      "distillation_loss": 0.04109591618180275,
      "epoch": 2.53,
      "learning_rate": 3.5543723591287916e-05,
      "loss": 0.0568,
      "step": 2667,
      "task_loss": 0.19810107350349426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7680821511442322,
      "compression_loss": 0.0,
      "distillation_loss": 0.03841260448098183,
      "epoch": 2.53,
      "learning_rate": 3.5534061387827936e-05,
      "loss": 0.0355,
      "step": 2668,
      "task_loss": 0.009714646264910698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7680987807976954,
      "compression_loss": 0.0,
      "distillation_loss": 0.11870106309652328,
      "epoch": 2.53,
      "learning_rate": 3.552439727080495e-05,
      "loss": 0.1244,
      "step": 2669,
      "task_loss": 0.1760600507259369
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7681154046739646,
      "compression_loss": 0.0,
      "distillation_loss": 0.10362593829631805,
      "epoch": 2.54,
      "learning_rate": 3.5514731241974544e-05,
      "loss": 0.1104,
      "step": 2670,
      "task_loss": 0.171173095703125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7681320227740432,
      "compression_loss": 0.0,
      "distillation_loss": 0.14736101031303406,
      "epoch": 2.54,
      "learning_rate": 3.5505063303092545e-05,
      "loss": 0.142,
      "step": 2671,
      "task_loss": 0.09334026277065277
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7681486350989352,
      "compression_loss": 0.0,
      "distillation_loss": 0.16119350492954254,
      "epoch": 2.54,
      "learning_rate": 3.549539345591521e-05,
      "loss": 0.1542,
      "step": 2672,
      "task_loss": 0.09096268564462662
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.768165241649644,
      "compression_loss": 0.0,
      "distillation_loss": 0.22717413306236267,
      "epoch": 2.54,
      "learning_rate": 3.5485721702199104e-05,
      "loss": 0.2191,
      "step": 2673,
      "task_loss": 0.1469288021326065
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7681818424271735,
      "compression_loss": 0.0,
      "distillation_loss": 0.04752913862466812,
      "epoch": 2.54,
      "learning_rate": 3.547604804370116e-05,
      "loss": 0.0435,
      "step": 2674,
      "task_loss": 0.007610570639371872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7681984374325271,
      "compression_loss": 0.0,
      "distillation_loss": 0.1687610149383545,
      "epoch": 2.54,
      "learning_rate": 3.5466372482178635e-05,
      "loss": 0.1833,
      "step": 2675,
      "task_loss": 0.31412065029144287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7682150266667088,
      "compression_loss": 0.0,
      "distillation_loss": 0.1348022222518921,
      "epoch": 2.54,
      "learning_rate": 3.545669501938913e-05,
      "loss": 0.1269,
      "step": 2676,
      "task_loss": 0.056100159883499146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7682316101307222,
      "compression_loss": 0.0,
      "distillation_loss": 0.20402267575263977,
      "epoch": 2.54,
      "learning_rate": 3.544701565709063e-05,
      "loss": 0.2063,
      "step": 2677,
      "task_loss": 0.22638899087905884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7682481878255708,
      "compression_loss": 0.0,
      "distillation_loss": 0.14998939633369446,
      "epoch": 2.54,
      "learning_rate": 3.54373343970414e-05,
      "loss": 0.143,
      "step": 2678,
      "task_loss": 0.08004312962293625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7682647597522585,
      "compression_loss": 0.0,
      "distillation_loss": 0.07659981399774551,
      "epoch": 2.54,
      "learning_rate": 3.542765124100014e-05,
      "loss": 0.0728,
      "step": 2679,
      "task_loss": 0.03887277841567993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7682813259117889,
      "compression_loss": 0.0,
      "distillation_loss": 0.06985659897327423,
      "epoch": 2.55,
      "learning_rate": 3.541796619072579e-05,
      "loss": 0.0637,
      "step": 2680,
      "task_loss": 0.008264170959591866
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7682978863051656,
      "compression_loss": 0.0,
      "distillation_loss": 0.14076358079910278,
      "epoch": 2.55,
      "learning_rate": 3.540827924797772e-05,
      "loss": 0.143,
      "step": 2681,
      "task_loss": 0.16294986009597778
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7683144409333924,
      "compression_loss": 0.0,
      "distillation_loss": 0.05743217468261719,
      "epoch": 2.55,
      "learning_rate": 3.5398590414515586e-05,
      "loss": 0.0619,
      "step": 2682,
      "task_loss": 0.10237649083137512
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7683309897974729,
      "compression_loss": 0.0,
      "distillation_loss": 0.0506620779633522,
      "epoch": 2.55,
      "learning_rate": 3.5388899692099433e-05,
      "loss": 0.0486,
      "step": 2683,
      "task_loss": 0.03040938824415207
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7683475328984108,
      "compression_loss": 0.0,
      "distillation_loss": 0.042747851461172104,
      "epoch": 2.55,
      "learning_rate": 3.537920708248961e-05,
      "loss": 0.062,
      "step": 2684,
      "task_loss": 0.23519286513328552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.76836407023721,
      "compression_loss": 0.0,
      "distillation_loss": 0.13504308462142944,
      "epoch": 2.55,
      "learning_rate": 3.536951258744684e-05,
      "loss": 0.1385,
      "step": 2685,
      "task_loss": 0.16998833417892456
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7683806018148738,
      "compression_loss": 0.0,
      "distillation_loss": 0.16051191091537476,
      "epoch": 2.55,
      "learning_rate": 3.5359816208732164e-05,
      "loss": 0.1578,
      "step": 2686,
      "task_loss": 0.13356226682662964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7683971276324061,
      "compression_loss": 0.0,
      "distillation_loss": 0.08552385866641998,
      "epoch": 2.55,
      "learning_rate": 3.535011794810698e-05,
      "loss": 0.0798,
      "step": 2687,
      "task_loss": 0.027825014665722847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7684136476908106,
      "compression_loss": 0.0,
      "distillation_loss": 0.04348837211728096,
      "epoch": 2.55,
      "learning_rate": 3.5340417807333026e-05,
      "loss": 0.0489,
      "step": 2688,
      "task_loss": 0.0971134603023529
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7684301619910909,
      "compression_loss": 0.0,
      "distillation_loss": 0.03343481570482254,
      "epoch": 2.55,
      "learning_rate": 3.533071578817239e-05,
      "loss": 0.0309,
      "step": 2689,
      "task_loss": 0.008328597992658615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7684466705342508,
      "compression_loss": 0.0,
      "distillation_loss": 0.09740258753299713,
      "epoch": 2.55,
      "learning_rate": 3.5321011892387455e-05,
      "loss": 0.0948,
      "step": 2690,
      "task_loss": 0.07093527913093567
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7684631733212938,
      "compression_loss": 0.0,
      "distillation_loss": 0.048360321670770645,
      "epoch": 2.56,
      "learning_rate": 3.5311306121741015e-05,
      "loss": 0.0522,
      "step": 2691,
      "task_loss": 0.08722103387117386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7684796703532236,
      "compression_loss": 0.0,
      "distillation_loss": 0.021398166194558144,
      "epoch": 2.56,
      "learning_rate": 3.530159847799616e-05,
      "loss": 0.0198,
      "step": 2692,
      "task_loss": 0.005231667309999466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7684961616310441,
      "compression_loss": 0.0,
      "distillation_loss": 0.2711659371852875,
      "epoch": 2.56,
      "learning_rate": 3.529188896291632e-05,
      "loss": 0.257,
      "step": 2693,
      "task_loss": 0.1298050582408905
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7685126471557587,
      "compression_loss": 0.0,
      "distillation_loss": 0.07550038397312164,
      "epoch": 2.56,
      "learning_rate": 3.5282177578265296e-05,
      "loss": 0.0708,
      "step": 2694,
      "task_loss": 0.028521571308374405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7685291269283714,
      "compression_loss": 0.0,
      "distillation_loss": 0.17867624759674072,
      "epoch": 2.56,
      "learning_rate": 3.527246432580718e-05,
      "loss": 0.1926,
      "step": 2695,
      "task_loss": 0.31801068782806396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7685456009498856,
      "compression_loss": 0.0,
      "distillation_loss": 0.1781276911497116,
      "epoch": 2.56,
      "learning_rate": 3.526274920730645e-05,
      "loss": 0.1767,
      "step": 2696,
      "task_loss": 0.16398294270038605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7685620692213051,
      "compression_loss": 0.0,
      "distillation_loss": 0.0298934206366539,
      "epoch": 2.56,
      "learning_rate": 3.525303222452791e-05,
      "loss": 0.0383,
      "step": 2697,
      "task_loss": 0.11393886804580688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7685785317436336,
      "compression_loss": 0.0,
      "distillation_loss": 0.03145752474665642,
      "epoch": 2.56,
      "learning_rate": 3.5243313379236685e-05,
      "loss": 0.0375,
      "step": 2698,
      "task_loss": 0.09145695716142654
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7685949885178748,
      "compression_loss": 0.0,
      "distillation_loss": 0.3614198565483093,
      "epoch": 2.56,
      "learning_rate": 3.5233592673198245e-05,
      "loss": 0.3515,
      "step": 2699,
      "task_loss": 0.26197120547294617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7686114395450322,
      "compression_loss": 0.0,
      "distillation_loss": 0.06908755749464035,
      "epoch": 2.56,
      "learning_rate": 3.522387010817842e-05,
      "loss": 0.0775,
      "step": 2700,
      "task_loss": 0.15282829105854034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7686278848261097,
      "compression_loss": 0.0,
      "distillation_loss": 0.07420150190591812,
      "epoch": 2.57,
      "learning_rate": 3.521414568594335e-05,
      "loss": 0.0918,
      "step": 2701,
      "task_loss": 0.25066766142845154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7686443243621108,
      "compression_loss": 0.0,
      "distillation_loss": 0.2168622463941574,
      "epoch": 2.57,
      "learning_rate": 3.520441940825952e-05,
      "loss": 0.2086,
      "step": 2702,
      "task_loss": 0.1346578747034073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7686607581540394,
      "compression_loss": 0.0,
      "distillation_loss": 0.2985137701034546,
      "epoch": 2.57,
      "learning_rate": 3.5194691276893755e-05,
      "loss": 0.2939,
      "step": 2703,
      "task_loss": 0.2523634433746338
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.768677186202899,
      "compression_loss": 0.0,
      "distillation_loss": 0.06703074276447296,
      "epoch": 2.57,
      "learning_rate": 3.518496129361323e-05,
      "loss": 0.0638,
      "step": 2704,
      "task_loss": 0.03458258882164955
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7686936085096935,
      "compression_loss": 0.0,
      "distillation_loss": 0.044759538024663925,
      "epoch": 2.57,
      "learning_rate": 3.5175229460185425e-05,
      "loss": 0.0411,
      "step": 2705,
      "task_loss": 0.007678527384996414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7687100250754263,
      "compression_loss": 0.0,
      "distillation_loss": 0.03616587817668915,
      "epoch": 2.57,
      "learning_rate": 3.5165495778378196e-05,
      "loss": 0.0336,
      "step": 2706,
      "task_loss": 0.010267160832881927
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7687264359011012,
      "compression_loss": 0.0,
      "distillation_loss": 0.22257006168365479,
      "epoch": 2.57,
      "learning_rate": 3.51557602499597e-05,
      "loss": 0.2122,
      "step": 2707,
      "task_loss": 0.1191282868385315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7687428409877219,
      "compression_loss": 0.0,
      "distillation_loss": 0.16167038679122925,
      "epoch": 2.57,
      "learning_rate": 3.514602287669844e-05,
      "loss": 0.1738,
      "step": 2708,
      "task_loss": 0.2829251289367676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7687592403362921,
      "compression_loss": 0.0,
      "distillation_loss": 0.020230792462825775,
      "epoch": 2.57,
      "learning_rate": 3.5136283660363255e-05,
      "loss": 0.0191,
      "step": 2709,
      "task_loss": 0.008950954303145409
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7687756339478155,
      "compression_loss": 0.0,
      "distillation_loss": 0.10566231608390808,
      "epoch": 2.57,
      "learning_rate": 3.5126542602723334e-05,
      "loss": 0.1088,
      "step": 2710,
      "task_loss": 0.13719259202480316
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7687920218232956,
      "compression_loss": 0.0,
      "distillation_loss": 0.16966648399829865,
      "epoch": 2.57,
      "learning_rate": 3.5116799705548175e-05,
      "loss": 0.1624,
      "step": 2711,
      "task_loss": 0.09719309955835342
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7688084039637363,
      "compression_loss": 0.0,
      "distillation_loss": 0.050816163420677185,
      "epoch": 2.58,
      "learning_rate": 3.510705497060762e-05,
      "loss": 0.0547,
      "step": 2712,
      "task_loss": 0.08981159329414368
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7688247803701412,
      "compression_loss": 0.0,
      "distillation_loss": 0.025936776772141457,
      "epoch": 2.58,
      "learning_rate": 3.509730839967187e-05,
      "loss": 0.0244,
      "step": 2713,
      "task_loss": 0.010662872344255447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.768841151043514,
      "compression_loss": 0.0,
      "distillation_loss": 0.07976531982421875,
      "epoch": 2.58,
      "learning_rate": 3.508755999451141e-05,
      "loss": 0.0739,
      "step": 2714,
      "task_loss": 0.020689811557531357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7688575159848584,
      "compression_loss": 0.0,
      "distillation_loss": 0.037483714520931244,
      "epoch": 2.58,
      "learning_rate": 3.50778097568971e-05,
      "loss": 0.04,
      "step": 2715,
      "task_loss": 0.06217388063669205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.768873875195178,
      "compression_loss": 0.0,
      "distillation_loss": 0.13560466468334198,
      "epoch": 2.58,
      "learning_rate": 3.506805768860011e-05,
      "loss": 0.1373,
      "step": 2716,
      "task_loss": 0.1521109640598297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7688902286754765,
      "compression_loss": 0.0,
      "distillation_loss": 0.0700618177652359,
      "epoch": 2.58,
      "learning_rate": 3.505830379139195e-05,
      "loss": 0.0776,
      "step": 2717,
      "task_loss": 0.1449713408946991
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7689065764267576,
      "compression_loss": 0.0,
      "distillation_loss": 0.1059018149971962,
      "epoch": 2.58,
      "learning_rate": 3.504854806704446e-05,
      "loss": 0.1082,
      "step": 2718,
      "task_loss": 0.12906883656978607
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.768922918450025,
      "compression_loss": 0.0,
      "distillation_loss": 0.06906574964523315,
      "epoch": 2.58,
      "learning_rate": 3.503879051732983e-05,
      "loss": 0.0792,
      "step": 2719,
      "task_loss": 0.16997401416301727
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7689392547462826,
      "compression_loss": 0.0,
      "distillation_loss": 0.03843502700328827,
      "epoch": 2.58,
      "learning_rate": 3.502903114402055e-05,
      "loss": 0.0541,
      "step": 2720,
      "task_loss": 0.1947551667690277
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7689555853165336,
      "compression_loss": 0.0,
      "distillation_loss": 0.07670299708843231,
      "epoch": 2.58,
      "learning_rate": 3.501926994888946e-05,
      "loss": 0.071,
      "step": 2721,
      "task_loss": 0.019545117393136024
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.768971910161782,
      "compression_loss": 0.0,
      "distillation_loss": 0.2852790057659149,
      "epoch": 2.58,
      "learning_rate": 3.500950693370974e-05,
      "loss": 0.2769,
      "step": 2722,
      "task_loss": 0.20112605392932892
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7689882292830315,
      "compression_loss": 0.0,
      "distillation_loss": 0.06108977273106575,
      "epoch": 2.59,
      "learning_rate": 3.499974210025487e-05,
      "loss": 0.0675,
      "step": 2723,
      "task_loss": 0.12486658245325089
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7690045426812856,
      "compression_loss": 0.0,
      "distillation_loss": 0.07272262871265411,
      "epoch": 2.59,
      "learning_rate": 3.4989975450298694e-05,
      "loss": 0.0792,
      "step": 2724,
      "task_loss": 0.13739514350891113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7690208503575482,
      "compression_loss": 0.0,
      "distillation_loss": 0.13614577054977417,
      "epoch": 2.59,
      "learning_rate": 3.498020698561536e-05,
      "loss": 0.1551,
      "step": 2725,
      "task_loss": 0.32593345642089844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7690371523128228,
      "compression_loss": 0.0,
      "distillation_loss": 0.0375310480594635,
      "epoch": 2.59,
      "learning_rate": 3.497043670797936e-05,
      "loss": 0.0366,
      "step": 2726,
      "task_loss": 0.028382275253534317
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7690534485481131,
      "compression_loss": 0.0,
      "distillation_loss": 0.04977725073695183,
      "epoch": 2.59,
      "learning_rate": 3.496066461916552e-05,
      "loss": 0.0467,
      "step": 2727,
      "task_loss": 0.019020110368728638
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7690697390644229,
      "compression_loss": 0.0,
      "distillation_loss": 0.1019824743270874,
      "epoch": 2.59,
      "learning_rate": 3.495089072094898e-05,
      "loss": 0.0978,
      "step": 2728,
      "task_loss": 0.06048191338777542
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7690860238627558,
      "compression_loss": 0.0,
      "distillation_loss": 0.021417532116174698,
      "epoch": 2.59,
      "learning_rate": 3.494111501510522e-05,
      "loss": 0.0293,
      "step": 2729,
      "task_loss": 0.10010670870542526
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7691023029441155,
      "compression_loss": 0.0,
      "distillation_loss": 0.04595291242003441,
      "epoch": 2.59,
      "learning_rate": 3.4931337503410034e-05,
      "loss": 0.0618,
      "step": 2730,
      "task_loss": 0.20489037036895752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7691185763095056,
      "compression_loss": 0.0,
      "distillation_loss": 0.07521625608205795,
      "epoch": 2.59,
      "learning_rate": 3.4921558187639556e-05,
      "loss": 0.0711,
      "step": 2731,
      "task_loss": 0.0338003970682621
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.76913484395993,
      "compression_loss": 0.0,
      "distillation_loss": 0.15600326657295227,
      "epoch": 2.59,
      "learning_rate": 3.491177706957026e-05,
      "loss": 0.1498,
      "step": 2732,
      "task_loss": 0.0938263088464737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7691511058963922,
      "compression_loss": 0.0,
      "distillation_loss": 0.19931530952453613,
      "epoch": 2.6,
      "learning_rate": 3.490199415097892e-05,
      "loss": 0.1947,
      "step": 2733,
      "task_loss": 0.1532270908355713
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.769167362119896,
      "compression_loss": 0.0,
      "distillation_loss": 0.08603714406490326,
      "epoch": 2.6,
      "learning_rate": 3.489220943364266e-05,
      "loss": 0.0839,
      "step": 2734,
      "task_loss": 0.06482724845409393
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7691836126314449,
      "compression_loss": 0.0,
      "distillation_loss": 0.0459892675280571,
      "epoch": 2.6,
      "learning_rate": 3.488242291933891e-05,
      "loss": 0.0517,
      "step": 2735,
      "task_loss": 0.10307285934686661
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7691998574320427,
      "compression_loss": 0.0,
      "distillation_loss": 0.09086348861455917,
      "epoch": 2.6,
      "learning_rate": 3.487263460984544e-05,
      "loss": 0.0902,
      "step": 2736,
      "task_loss": 0.08435454964637756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7692160965226931,
      "compression_loss": 0.0,
      "distillation_loss": 0.14168088138103485,
      "epoch": 2.6,
      "learning_rate": 3.486284450694035e-05,
      "loss": 0.1393,
      "step": 2737,
      "task_loss": 0.11769342422485352
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7692323299043997,
      "compression_loss": 0.0,
      "distillation_loss": 0.06477680802345276,
      "epoch": 2.6,
      "learning_rate": 3.485305261240205e-05,
      "loss": 0.0647,
      "step": 2738,
      "task_loss": 0.06358174979686737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7692485575781663,
      "compression_loss": 0.0,
      "distillation_loss": 0.10200376808643341,
      "epoch": 2.6,
      "learning_rate": 3.4843258928009294e-05,
      "loss": 0.1103,
      "step": 2739,
      "task_loss": 0.18466830253601074
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7692647795449965,
      "compression_loss": 0.0,
      "distillation_loss": 0.12464478611946106,
      "epoch": 2.6,
      "learning_rate": 3.4833463455541146e-05,
      "loss": 0.1179,
      "step": 2740,
      "task_loss": 0.05756617337465286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.769280995805894,
      "compression_loss": 0.0,
      "distillation_loss": 0.38097622990608215,
      "epoch": 2.6,
      "learning_rate": 3.4823666196777006e-05,
      "loss": 0.3663,
      "step": 2741,
      "task_loss": 0.23438085615634918
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7692972063618625,
      "compression_loss": 0.0,
      "distillation_loss": 0.09649477154016495,
      "epoch": 2.6,
      "learning_rate": 3.481386715349659e-05,
      "loss": 0.0907,
      "step": 2742,
      "task_loss": 0.03846879303455353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7693134112139056,
      "compression_loss": 0.0,
      "distillation_loss": 0.049957577139139175,
      "epoch": 2.6,
      "learning_rate": 3.4804066327479934e-05,
      "loss": 0.0601,
      "step": 2743,
      "task_loss": 0.15115505456924438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7693296103630272,
      "compression_loss": 0.0,
      "distillation_loss": 0.06845077127218246,
      "epoch": 2.61,
      "learning_rate": 3.4794263720507427e-05,
      "loss": 0.071,
      "step": 2744,
      "task_loss": 0.09412077814340591
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7693458038102307,
      "compression_loss": 0.0,
      "distillation_loss": 0.09953967481851578,
      "epoch": 2.61,
      "learning_rate": 3.478445933435973e-05,
      "loss": 0.1021,
      "step": 2745,
      "task_loss": 0.1255946159362793
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.76936199155652,
      "compression_loss": 0.0,
      "distillation_loss": 0.04850146174430847,
      "epoch": 2.61,
      "learning_rate": 3.4774653170817884e-05,
      "loss": 0.0495,
      "step": 2746,
      "task_loss": 0.0581977404654026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7693781736028985,
      "compression_loss": 0.0,
      "distillation_loss": 0.06344247609376907,
      "epoch": 2.61,
      "learning_rate": 3.4764845231663205e-05,
      "loss": 0.0654,
      "step": 2747,
      "task_loss": 0.0828084945678711
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7693943499503703,
      "compression_loss": 0.0,
      "distillation_loss": 0.16078636050224304,
      "epoch": 2.61,
      "learning_rate": 3.475503551867737e-05,
      "loss": 0.1636,
      "step": 2748,
      "task_loss": 0.1886562705039978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7694105205999388,
      "compression_loss": 0.0,
      "distillation_loss": 0.12109316885471344,
      "epoch": 2.61,
      "learning_rate": 3.474522403364235e-05,
      "loss": 0.1212,
      "step": 2749,
      "task_loss": 0.12180116772651672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7694266855526076,
      "compression_loss": 0.0,
      "distillation_loss": 0.11483351141214371,
      "epoch": 2.61,
      "learning_rate": 3.473541077834045e-05,
      "loss": 0.1184,
      "step": 2750,
      "task_loss": 0.1501813381910324
    },
    {
      "epoch": 2.61,
      "eval_accuracy": 0.893348623853211,
      "eval_loss": 0.4236961901187897,
      "eval_runtime": 18.2679,
      "eval_samples_per_second": 47.734,
      "eval_steps_per_second": 5.967,
      "step": 2750
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7694428448093806,
      "compression_loss": 0.0,
      "distillation_loss": 0.11966343224048615,
      "epoch": 2.61,
      "learning_rate": 3.4725595754554295e-05,
      "loss": 0.1252,
      "step": 2751,
      "task_loss": 0.17513278126716614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7694589983712615,
      "compression_loss": 0.0,
      "distillation_loss": 0.10641565918922424,
      "epoch": 2.61,
      "learning_rate": 3.471577896406683e-05,
      "loss": 0.1126,
      "step": 2752,
      "task_loss": 0.1684948056936264
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7694751462392538,
      "compression_loss": 0.0,
      "distillation_loss": 0.08618808537721634,
      "epoch": 2.61,
      "learning_rate": 3.470596040866133e-05,
      "loss": 0.0846,
      "step": 2753,
      "task_loss": 0.07057564705610275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7694912884143613,
      "compression_loss": 0.0,
      "distillation_loss": 0.05224674940109253,
      "epoch": 2.62,
      "learning_rate": 3.4696140090121376e-05,
      "loss": 0.0487,
      "step": 2754,
      "task_loss": 0.016593074426054955
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7695074248975876,
      "compression_loss": 0.0,
      "distillation_loss": 0.1727830171585083,
      "epoch": 2.62,
      "learning_rate": 3.468631801023088e-05,
      "loss": 0.168,
      "step": 2755,
      "task_loss": 0.12460935115814209
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7695235556899365,
      "compression_loss": 0.0,
      "distillation_loss": 0.04192081093788147,
      "epoch": 2.62,
      "learning_rate": 3.467649417077406e-05,
      "loss": 0.0526,
      "step": 2756,
      "task_loss": 0.14908772706985474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7695396807924116,
      "compression_loss": 0.0,
      "distillation_loss": 0.13823646306991577,
      "epoch": 2.62,
      "learning_rate": 3.466666857353547e-05,
      "loss": 0.1403,
      "step": 2757,
      "task_loss": 0.15889990329742432
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7695558002060165,
      "compression_loss": 0.0,
      "distillation_loss": 0.08324067294597626,
      "epoch": 2.62,
      "learning_rate": 3.465684122029999e-05,
      "loss": 0.0865,
      "step": 2758,
      "task_loss": 0.11616663634777069
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7695719139317552,
      "compression_loss": 0.0,
      "distillation_loss": 0.18361113965511322,
      "epoch": 2.62,
      "learning_rate": 3.464701211285279e-05,
      "loss": 0.1828,
      "step": 2759,
      "task_loss": 0.1751021444797516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7695880219706309,
      "compression_loss": 0.0,
      "distillation_loss": 0.038572054356336594,
      "epoch": 2.62,
      "learning_rate": 3.463718125297937e-05,
      "loss": 0.0367,
      "step": 2760,
      "task_loss": 0.020345257595181465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7696041243236478,
      "compression_loss": 0.0,
      "distillation_loss": 0.11891628056764603,
      "epoch": 2.62,
      "learning_rate": 3.462734864246557e-05,
      "loss": 0.1142,
      "step": 2761,
      "task_loss": 0.07189692556858063
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7696202209918092,
      "compression_loss": 0.0,
      "distillation_loss": 0.0748853087425232,
      "epoch": 2.62,
      "learning_rate": 3.4617514283097524e-05,
      "loss": 0.0825,
      "step": 2762,
      "task_loss": 0.1514107882976532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7696363119761188,
      "compression_loss": 0.0,
      "distillation_loss": 0.17334090173244476,
      "epoch": 2.62,
      "learning_rate": 3.4607678176661695e-05,
      "loss": 0.1803,
      "step": 2763,
      "task_loss": 0.2433452010154724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7696523972775805,
      "compression_loss": 0.0,
      "distillation_loss": 0.08427559584379196,
      "epoch": 2.62,
      "learning_rate": 3.459784032494484e-05,
      "loss": 0.1008,
      "step": 2764,
      "task_loss": 0.24964390695095062
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7696684768971979,
      "compression_loss": 0.0,
      "distillation_loss": 0.04236322641372681,
      "epoch": 2.63,
      "learning_rate": 3.458800072973408e-05,
      "loss": 0.0393,
      "step": 2765,
      "task_loss": 0.011962385848164558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7696845508359745,
      "compression_loss": 0.0,
      "distillation_loss": 0.04558961093425751,
      "epoch": 2.63,
      "learning_rate": 3.45781593928168e-05,
      "loss": 0.051,
      "step": 2766,
      "task_loss": 0.09958979487419128
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7697006190949143,
      "compression_loss": 0.0,
      "distillation_loss": 0.05856577679514885,
      "epoch": 2.63,
      "learning_rate": 3.4568316315980745e-05,
      "loss": 0.0557,
      "step": 2767,
      "task_loss": 0.029869040474295616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7697166816750208,
      "compression_loss": 0.0,
      "distillation_loss": 0.14246824383735657,
      "epoch": 2.63,
      "learning_rate": 3.455847150101395e-05,
      "loss": 0.1396,
      "step": 2768,
      "task_loss": 0.11401065438985825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7697327385772976,
      "compression_loss": 0.0,
      "distillation_loss": 0.08908317983150482,
      "epoch": 2.63,
      "learning_rate": 3.454862494970477e-05,
      "loss": 0.1026,
      "step": 2769,
      "task_loss": 0.22423215210437775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7697487898027486,
      "compression_loss": 0.0,
      "distillation_loss": 0.11774835735559464,
      "epoch": 2.63,
      "learning_rate": 3.4538776663841875e-05,
      "loss": 0.1259,
      "step": 2770,
      "task_loss": 0.19950447976589203
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7697648353523773,
      "compression_loss": 0.0,
      "distillation_loss": 0.07487460970878601,
      "epoch": 2.63,
      "learning_rate": 3.452892664521427e-05,
      "loss": 0.0828,
      "step": 2771,
      "task_loss": 0.15448282659053802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7697808752271874,
      "compression_loss": 0.0,
      "distillation_loss": 0.13396842777729034,
      "epoch": 2.63,
      "learning_rate": 3.4519074895611244e-05,
      "loss": 0.1337,
      "step": 2772,
      "task_loss": 0.13105513155460358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7697969094281827,
      "compression_loss": 0.0,
      "distillation_loss": 0.03769555687904358,
      "epoch": 2.63,
      "learning_rate": 3.4509221416822415e-05,
      "loss": 0.0348,
      "step": 2773,
      "task_loss": 0.00889505073428154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7698129379563668,
      "compression_loss": 0.0,
      "distillation_loss": 0.1526089906692505,
      "epoch": 2.63,
      "learning_rate": 3.4499366210637725e-05,
      "loss": 0.1464,
      "step": 2774,
      "task_loss": 0.09016816318035126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7698289608127434,
      "compression_loss": 0.0,
      "distillation_loss": 0.05517404526472092,
      "epoch": 2.64,
      "learning_rate": 3.4489509278847414e-05,
      "loss": 0.0691,
      "step": 2775,
      "task_loss": 0.19403240084648132
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7698449779983162,
      "compression_loss": 0.0,
      "distillation_loss": 0.14141535758972168,
      "epoch": 2.64,
      "learning_rate": 3.4479650623242036e-05,
      "loss": 0.1409,
      "step": 2776,
      "task_loss": 0.13665583729743958
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7698609895140888,
      "compression_loss": 0.0,
      "distillation_loss": 0.1828983724117279,
      "epoch": 2.64,
      "learning_rate": 3.446979024561246e-05,
      "loss": 0.1816,
      "step": 2777,
      "task_loss": 0.1703716665506363
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7698769953610649,
      "compression_loss": 0.0,
      "distillation_loss": 0.031704775989055634,
      "epoch": 2.64,
      "learning_rate": 3.44599281477499e-05,
      "loss": 0.0384,
      "step": 2778,
      "task_loss": 0.098890021443367
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7698929955402484,
      "compression_loss": 0.0,
      "distillation_loss": 0.03340402618050575,
      "epoch": 2.64,
      "learning_rate": 3.4450064331445814e-05,
      "loss": 0.0319,
      "step": 2779,
      "task_loss": 0.018638523295521736
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7699089900526426,
      "compression_loss": 0.0,
      "distillation_loss": 0.040086835622787476,
      "epoch": 2.64,
      "learning_rate": 3.444019879849204e-05,
      "loss": 0.048,
      "step": 2780,
      "task_loss": 0.11879396438598633
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7699249788992515,
      "compression_loss": 0.0,
      "distillation_loss": 0.05169200897216797,
      "epoch": 2.64,
      "learning_rate": 3.443033155068069e-05,
      "loss": 0.0542,
      "step": 2781,
      "task_loss": 0.07657624781131744
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7699409620810785,
      "compression_loss": 0.0,
      "distillation_loss": 0.20420724153518677,
      "epoch": 2.64,
      "learning_rate": 3.442046258980419e-05,
      "loss": 0.1945,
      "step": 2782,
      "task_loss": 0.10754656046628952
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7699569395991277,
      "compression_loss": 0.0,
      "distillation_loss": 0.1341618299484253,
      "epoch": 2.64,
      "learning_rate": 3.4410591917655296e-05,
      "loss": 0.1326,
      "step": 2783,
      "task_loss": 0.1190015971660614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7699729114544024,
      "compression_loss": 0.0,
      "distillation_loss": 0.09340974688529968,
      "epoch": 2.64,
      "learning_rate": 3.4400719536027056e-05,
      "loss": 0.0875,
      "step": 2784,
      "task_loss": 0.03399471938610077
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7699888776479064,
      "compression_loss": 0.0,
      "distillation_loss": 0.19588154554367065,
      "epoch": 2.64,
      "learning_rate": 3.4390845446712836e-05,
      "loss": 0.187,
      "step": 2785,
      "task_loss": 0.10719942301511765
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7700048381806435,
      "compression_loss": 0.0,
      "distillation_loss": 0.11635538935661316,
      "epoch": 2.65,
      "learning_rate": 3.438096965150632e-05,
      "loss": 0.1287,
      "step": 2786,
      "task_loss": 0.2399107664823532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7700207930536171,
      "compression_loss": 0.0,
      "distillation_loss": 0.17617712914943695,
      "epoch": 2.65,
      "learning_rate": 3.4371092152201485e-05,
      "loss": 0.1735,
      "step": 2787,
      "task_loss": 0.14985665678977966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7700367422678311,
      "compression_loss": 0.0,
      "distillation_loss": 0.06373878568410873,
      "epoch": 2.65,
      "learning_rate": 3.4361212950592624e-05,
      "loss": 0.0669,
      "step": 2788,
      "task_loss": 0.09571292251348495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7700526858242893,
      "compression_loss": 0.0,
      "distillation_loss": 0.05058780312538147,
      "epoch": 2.65,
      "learning_rate": 3.435133204847435e-05,
      "loss": 0.0469,
      "step": 2789,
      "task_loss": 0.013910744339227676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7700686237239951,
      "compression_loss": 0.0,
      "distillation_loss": 0.10628429800271988,
      "epoch": 2.65,
      "learning_rate": 3.4341449447641575e-05,
      "loss": 0.1091,
      "step": 2790,
      "task_loss": 0.13484588265419006
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7700845559679523,
      "compression_loss": 0.0,
      "distillation_loss": 0.024575646966695786,
      "epoch": 2.65,
      "learning_rate": 3.433156514988951e-05,
      "loss": 0.0287,
      "step": 2791,
      "task_loss": 0.06540574878454208
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7701004825571647,
      "compression_loss": 0.0,
      "distillation_loss": 0.11763833463191986,
      "epoch": 2.65,
      "learning_rate": 3.432167915701371e-05,
      "loss": 0.1207,
      "step": 2792,
      "task_loss": 0.14855128526687622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7701164034926358,
      "compression_loss": 0.0,
      "distillation_loss": 0.04107709601521492,
      "epoch": 2.65,
      "learning_rate": 3.431179147080999e-05,
      "loss": 0.0611,
      "step": 2793,
      "task_loss": 0.24124276638031006
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7701323187753693,
      "compression_loss": 0.0,
      "distillation_loss": 0.03191131353378296,
      "epoch": 2.65,
      "learning_rate": 3.4301902093074504e-05,
      "loss": 0.0301,
      "step": 2794,
      "task_loss": 0.013331804424524307
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.770148228406369,
      "compression_loss": 0.0,
      "distillation_loss": 0.029843294993042946,
      "epoch": 2.65,
      "learning_rate": 3.42920110256037e-05,
      "loss": 0.0277,
      "step": 2795,
      "task_loss": 0.008593736216425896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7701641323866384,
      "compression_loss": 0.0,
      "distillation_loss": 0.05394424870610237,
      "epoch": 2.66,
      "learning_rate": 3.428211827019434e-05,
      "loss": 0.0504,
      "step": 2796,
      "task_loss": 0.018413875252008438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7701800307171814,
      "compression_loss": 0.0,
      "distillation_loss": 0.12455835193395615,
      "epoch": 2.66,
      "learning_rate": 3.42722238286435e-05,
      "loss": 0.1222,
      "step": 2797,
      "task_loss": 0.10132066905498505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7701959233990016,
      "compression_loss": 0.0,
      "distillation_loss": 0.036928869783878326,
      "epoch": 2.66,
      "learning_rate": 3.426232770274855e-05,
      "loss": 0.0458,
      "step": 2798,
      "task_loss": 0.12519891560077667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7702118104331026,
      "compression_loss": 0.0,
      "distillation_loss": 0.03775065019726753,
      "epoch": 2.66,
      "learning_rate": 3.4252429894307154e-05,
      "loss": 0.0426,
      "step": 2799,
      "task_loss": 0.08613856136798859
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7702276918204881,
      "compression_loss": 0.0,
      "distillation_loss": 0.14951038360595703,
      "epoch": 2.66,
      "learning_rate": 3.424253040511731e-05,
      "loss": 0.1439,
      "step": 2800,
      "task_loss": 0.09322544187307358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.770243567562162,
      "compression_loss": 0.0,
      "distillation_loss": 0.14305658638477325,
      "epoch": 2.66,
      "learning_rate": 3.4232629236977316e-05,
      "loss": 0.1513,
      "step": 2801,
      "task_loss": 0.2257470041513443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7702594376591276,
      "compression_loss": 0.0,
      "distillation_loss": 0.08632157742977142,
      "epoch": 2.66,
      "learning_rate": 3.4222726391685746e-05,
      "loss": 0.0854,
      "step": 2802,
      "task_loss": 0.07660988718271255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7702753021123889,
      "compression_loss": 0.0,
      "distillation_loss": 0.12226057052612305,
      "epoch": 2.66,
      "learning_rate": 3.42128218710415e-05,
      "loss": 0.1323,
      "step": 2803,
      "task_loss": 0.22227300703525543
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7702911609229496,
      "compression_loss": 0.0,
      "distillation_loss": 0.05076700448989868,
      "epoch": 2.66,
      "learning_rate": 3.420291567684381e-05,
      "loss": 0.0488,
      "step": 2804,
      "task_loss": 0.03137904405593872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7703070140918131,
      "compression_loss": 0.0,
      "distillation_loss": 0.13917645812034607,
      "epoch": 2.66,
      "learning_rate": 3.419300781089216e-05,
      "loss": 0.1431,
      "step": 2805,
      "task_loss": 0.17885732650756836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7703228616199832,
      "compression_loss": 0.0,
      "distillation_loss": 0.047890555113554,
      "epoch": 2.66,
      "learning_rate": 3.418309827498637e-05,
      "loss": 0.053,
      "step": 2806,
      "task_loss": 0.09856240451335907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7703387035084637,
      "compression_loss": 0.0,
      "distillation_loss": 0.019215038046240807,
      "epoch": 2.67,
      "learning_rate": 3.4173187070926546e-05,
      "loss": 0.0224,
      "step": 2807,
      "task_loss": 0.05135765299201012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7703545397582582,
      "compression_loss": 0.0,
      "distillation_loss": 0.07260240614414215,
      "epoch": 2.67,
      "learning_rate": 3.4163274200513116e-05,
      "loss": 0.0875,
      "step": 2808,
      "task_loss": 0.221457377076149
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7703703703703704,
      "compression_loss": 0.0,
      "distillation_loss": 0.06960202753543854,
      "epoch": 2.67,
      "learning_rate": 3.415335966554679e-05,
      "loss": 0.0853,
      "step": 2809,
      "task_loss": 0.22612322866916656
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7703861953458039,
      "compression_loss": 0.0,
      "distillation_loss": 0.0772671177983284,
      "epoch": 2.67,
      "learning_rate": 3.414344346782861e-05,
      "loss": 0.0818,
      "step": 2810,
      "task_loss": 0.12265747040510178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7704020146855625,
      "compression_loss": 0.0,
      "distillation_loss": 0.16112419962882996,
      "epoch": 2.67,
      "learning_rate": 3.413352560915988e-05,
      "loss": 0.1535,
      "step": 2811,
      "task_loss": 0.0847577229142189
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7704178283906499,
      "compression_loss": 0.0,
      "distillation_loss": 0.23741042613983154,
      "epoch": 2.67,
      "learning_rate": 3.412360609134223e-05,
      "loss": 0.2338,
      "step": 2812,
      "task_loss": 0.20153185725212097
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7704336364620695,
      "compression_loss": 0.0,
      "distillation_loss": 0.10814424604177475,
      "epoch": 2.67,
      "learning_rate": 3.411368491617761e-05,
      "loss": 0.1036,
      "step": 2813,
      "task_loss": 0.06266738474369049
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7704494389008254,
      "compression_loss": 0.0,
      "distillation_loss": 0.06540797650814056,
      "epoch": 2.67,
      "learning_rate": 3.410376208546822e-05,
      "loss": 0.0606,
      "step": 2814,
      "task_loss": 0.017390571534633636
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7704652357079209,
      "compression_loss": 0.0,
      "distillation_loss": 0.033819716423749924,
      "epoch": 2.67,
      "learning_rate": 3.409383760101661e-05,
      "loss": 0.0382,
      "step": 2815,
      "task_loss": 0.07716642320156097
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.77048102688436,
      "compression_loss": 0.0,
      "distillation_loss": 0.20797955989837646,
      "epoch": 2.67,
      "learning_rate": 3.4083911464625596e-05,
      "loss": 0.1993,
      "step": 2816,
      "task_loss": 0.12118206918239594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7704968124311462,
      "compression_loss": 0.0,
      "distillation_loss": 0.13762623071670532,
      "epoch": 2.68,
      "learning_rate": 3.407398367809832e-05,
      "loss": 0.1356,
      "step": 2817,
      "task_loss": 0.11715132743120193
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7705125923492833,
      "compression_loss": 0.0,
      "distillation_loss": 0.14595434069633484,
      "epoch": 2.68,
      "learning_rate": 3.406405424323821e-05,
      "loss": 0.1374,
      "step": 2818,
      "task_loss": 0.06045624241232872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7705283666397748,
      "compression_loss": 0.0,
      "distillation_loss": 0.030591927468776703,
      "epoch": 2.68,
      "learning_rate": 3.4054123161848995e-05,
      "loss": 0.0285,
      "step": 2819,
      "task_loss": 0.010143185034394264
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7705441353036244,
      "compression_loss": 0.0,
      "distillation_loss": 0.21557293832302094,
      "epoch": 2.68,
      "learning_rate": 3.4044190435734695e-05,
      "loss": 0.2094,
      "step": 2820,
      "task_loss": 0.15403705835342407
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.770559898341836,
      "compression_loss": 0.0,
      "distillation_loss": 0.04488247632980347,
      "epoch": 2.68,
      "learning_rate": 3.403425606669965e-05,
      "loss": 0.0427,
      "step": 2821,
      "task_loss": 0.023530708625912666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7705756557554131,
      "compression_loss": 0.0,
      "distillation_loss": 0.1816938817501068,
      "epoch": 2.68,
      "learning_rate": 3.4024320056548475e-05,
      "loss": 0.174,
      "step": 2822,
      "task_loss": 0.10455642640590668
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7705914075453595,
      "compression_loss": 0.0,
      "distillation_loss": 0.03927179425954819,
      "epoch": 2.68,
      "learning_rate": 3.401438240708611e-05,
      "loss": 0.0501,
      "step": 2823,
      "task_loss": 0.14788560569286346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7706071537126787,
      "compression_loss": 0.0,
      "distillation_loss": 0.1478666365146637,
      "epoch": 2.68,
      "learning_rate": 3.400444312011776e-05,
      "loss": 0.1539,
      "step": 2824,
      "task_loss": 0.20857636630535126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7706228942583746,
      "compression_loss": 0.0,
      "distillation_loss": 0.1341654360294342,
      "epoch": 2.68,
      "learning_rate": 3.399450219744896e-05,
      "loss": 0.1379,
      "step": 2825,
      "task_loss": 0.17152555286884308
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7706386291834507,
      "compression_loss": 0.0,
      "distillation_loss": 0.09930291771888733,
      "epoch": 2.68,
      "learning_rate": 3.3984559640885505e-05,
      "loss": 0.1029,
      "step": 2826,
      "task_loss": 0.1356475055217743
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7706543584889108,
      "compression_loss": 0.0,
      "distillation_loss": 0.049742672592401505,
      "epoch": 2.68,
      "learning_rate": 3.3974615452233526e-05,
      "loss": 0.0463,
      "step": 2827,
      "task_loss": 0.015741702169179916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7706700821757586,
      "compression_loss": 0.0,
      "distillation_loss": 0.047701530158519745,
      "epoch": 2.69,
      "learning_rate": 3.396466963329944e-05,
      "loss": 0.0506,
      "step": 2828,
      "task_loss": 0.0768967792391777
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7706858002449977,
      "compression_loss": 0.0,
      "distillation_loss": 0.14289794862270355,
      "epoch": 2.69,
      "learning_rate": 3.395472218588992e-05,
      "loss": 0.1379,
      "step": 2829,
      "task_loss": 0.09291721880435944
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7707015126976318,
      "compression_loss": 0.0,
      "distillation_loss": 0.11319032311439514,
      "epoch": 2.69,
      "learning_rate": 3.394477311181201e-05,
      "loss": 0.1067,
      "step": 2830,
      "task_loss": 0.04876325652003288
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7707172195346645,
      "compression_loss": 0.0,
      "distillation_loss": 0.14416098594665527,
      "epoch": 2.69,
      "learning_rate": 3.393482241287297e-05,
      "loss": 0.1571,
      "step": 2831,
      "task_loss": 0.2737889587879181
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7707329207570998,
      "compression_loss": 0.0,
      "distillation_loss": 0.1706182062625885,
      "epoch": 2.69,
      "learning_rate": 3.392487009088041e-05,
      "loss": 0.1614,
      "step": 2832,
      "task_loss": 0.07805980741977692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7707486163659409,
      "compression_loss": 0.0,
      "distillation_loss": 0.14137789607048035,
      "epoch": 2.69,
      "learning_rate": 3.391491614764222e-05,
      "loss": 0.1342,
      "step": 2833,
      "task_loss": 0.07003333419561386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7707643063621918,
      "compression_loss": 0.0,
      "distillation_loss": 0.09552104771137238,
      "epoch": 2.69,
      "learning_rate": 3.390496058496657e-05,
      "loss": 0.1013,
      "step": 2834,
      "task_loss": 0.15300557017326355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7707799907468562,
      "compression_loss": 0.0,
      "distillation_loss": 0.1020299419760704,
      "epoch": 2.69,
      "learning_rate": 3.3895003404661955e-05,
      "loss": 0.1235,
      "step": 2835,
      "task_loss": 0.3162316679954529
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7707956695209377,
      "compression_loss": 0.0,
      "distillation_loss": 0.10217801481485367,
      "epoch": 2.69,
      "learning_rate": 3.3885044608537125e-05,
      "loss": 0.1284,
      "step": 2836,
      "task_loss": 0.3640749156475067
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7708113426854399,
      "compression_loss": 0.0,
      "distillation_loss": 0.09118107706308365,
      "epoch": 2.69,
      "learning_rate": 3.387508419840115e-05,
      "loss": 0.0858,
      "step": 2837,
      "task_loss": 0.03702104091644287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7708270102413666,
      "compression_loss": 0.0,
      "distillation_loss": 0.123194620013237,
      "epoch": 2.7,
      "learning_rate": 3.386512217606339e-05,
      "loss": 0.1161,
      "step": 2838,
      "task_loss": 0.05249645560979843
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7708426721897215,
      "compression_loss": 0.0,
      "distillation_loss": 0.0807013064622879,
      "epoch": 2.7,
      "learning_rate": 3.385515854333349e-05,
      "loss": 0.0786,
      "step": 2839,
      "task_loss": 0.06016937270760536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7708583285315082,
      "compression_loss": 0.0,
      "distillation_loss": 0.08974245935678482,
      "epoch": 2.7,
      "learning_rate": 3.38451933020214e-05,
      "loss": 0.0873,
      "step": 2840,
      "task_loss": 0.0655125230550766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7708739792677303,
      "compression_loss": 0.0,
      "distillation_loss": 0.18268810212612152,
      "epoch": 2.7,
      "learning_rate": 3.383522645393734e-05,
      "loss": 0.1749,
      "step": 2841,
      "task_loss": 0.10475851595401764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7708896243993917,
      "compression_loss": 0.0,
      "distillation_loss": 0.23169159889221191,
      "epoch": 2.7,
      "learning_rate": 3.3825258000891846e-05,
      "loss": 0.2429,
      "step": 2842,
      "task_loss": 0.34396815299987793
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7709052639274959,
      "compression_loss": 0.0,
      "distillation_loss": 0.13757774233818054,
      "epoch": 2.7,
      "learning_rate": 3.381528794469574e-05,
      "loss": 0.1312,
      "step": 2843,
      "task_loss": 0.07409615814685822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7709208978530467,
      "compression_loss": 0.0,
      "distillation_loss": 0.029491236433386803,
      "epoch": 2.7,
      "learning_rate": 3.380531628716012e-05,
      "loss": 0.0344,
      "step": 2844,
      "task_loss": 0.07904480397701263
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7709365261770478,
      "compression_loss": 0.0,
      "distillation_loss": 0.08474864065647125,
      "epoch": 2.7,
      "learning_rate": 3.3795343030096384e-05,
      "loss": 0.0934,
      "step": 2845,
      "task_loss": 0.17167925834655762
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7709521489005028,
      "compression_loss": 0.0,
      "distillation_loss": 0.03882834315299988,
      "epoch": 2.7,
      "learning_rate": 3.3785368175316226e-05,
      "loss": 0.0392,
      "step": 2846,
      "task_loss": 0.042411088943481445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7709677660244153,
      "compression_loss": 0.0,
      "distillation_loss": 0.04816707968711853,
      "epoch": 2.7,
      "learning_rate": 3.377539172463164e-05,
      "loss": 0.0444,
      "step": 2847,
      "task_loss": 0.010003169998526573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7709833775497892,
      "compression_loss": 0.0,
      "distillation_loss": 0.09748025238513947,
      "epoch": 2.7,
      "learning_rate": 3.376541367985488e-05,
      "loss": 0.1011,
      "step": 2848,
      "task_loss": 0.13417553901672363
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.770998983477628,
      "compression_loss": 0.0,
      "distillation_loss": 0.032245099544525146,
      "epoch": 2.71,
      "learning_rate": 3.3755434042798506e-05,
      "loss": 0.0298,
      "step": 2849,
      "task_loss": 0.008108781650662422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7710145838089355,
      "compression_loss": 0.0,
      "distillation_loss": 0.028734374791383743,
      "epoch": 2.71,
      "learning_rate": 3.374545281527538e-05,
      "loss": 0.0302,
      "step": 2850,
      "task_loss": 0.043056122958660126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7710301785447152,
      "compression_loss": 0.0,
      "distillation_loss": 0.04646429419517517,
      "epoch": 2.71,
      "learning_rate": 3.3735469999098615e-05,
      "loss": 0.0477,
      "step": 2851,
      "task_loss": 0.05848044529557228
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7710457676859711,
      "compression_loss": 0.0,
      "distillation_loss": 0.09431092441082001,
      "epoch": 2.71,
      "learning_rate": 3.372548559608166e-05,
      "loss": 0.101,
      "step": 2852,
      "task_loss": 0.16074809432029724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7710613512337066,
      "compression_loss": 0.0,
      "distillation_loss": 0.042417608201503754,
      "epoch": 2.71,
      "learning_rate": 3.3715499608038234e-05,
      "loss": 0.0391,
      "step": 2853,
      "task_loss": 0.009729046374559402
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7710769291889255,
      "compression_loss": 0.0,
      "distillation_loss": 0.10295215994119644,
      "epoch": 2.71,
      "learning_rate": 3.370551203678233e-05,
      "loss": 0.0966,
      "step": 2854,
      "task_loss": 0.03992331400513649
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7710925015526314,
      "compression_loss": 0.0,
      "distillation_loss": 0.026899177581071854,
      "epoch": 2.71,
      "learning_rate": 3.369552288412822e-05,
      "loss": 0.0407,
      "step": 2855,
      "task_loss": 0.1650460809469223
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7711080683258281,
      "compression_loss": 0.0,
      "distillation_loss": 0.09420739114284515,
      "epoch": 2.71,
      "learning_rate": 3.368553215189052e-05,
      "loss": 0.0965,
      "step": 2856,
      "task_loss": 0.1171763613820076
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7711236295095193,
      "compression_loss": 0.0,
      "distillation_loss": 0.0723368227481842,
      "epoch": 2.71,
      "learning_rate": 3.367553984188407e-05,
      "loss": 0.0738,
      "step": 2857,
      "task_loss": 0.08675511181354523
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7711391851047085,
      "compression_loss": 0.0,
      "distillation_loss": 0.10755441337823868,
      "epoch": 2.71,
      "learning_rate": 3.366554595592402e-05,
      "loss": 0.103,
      "step": 2858,
      "task_loss": 0.06218289956450462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7711547351123995,
      "compression_loss": 0.0,
      "distillation_loss": 0.11410737037658691,
      "epoch": 2.72,
      "learning_rate": 3.365555049582582e-05,
      "loss": 0.1191,
      "step": 2859,
      "task_loss": 0.16380539536476135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.771170279533596,
      "compression_loss": 0.0,
      "distillation_loss": 0.09805843234062195,
      "epoch": 2.72,
      "learning_rate": 3.364555346340518e-05,
      "loss": 0.0936,
      "step": 2860,
      "task_loss": 0.053569987416267395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7711858183693016,
      "compression_loss": 0.0,
      "distillation_loss": 0.11211707442998886,
      "epoch": 2.72,
      "learning_rate": 3.3635554860478126e-05,
      "loss": 0.1062,
      "step": 2861,
      "task_loss": 0.053317755460739136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7712013516205201,
      "compression_loss": 0.0,
      "distillation_loss": 0.06892388314008713,
      "epoch": 2.72,
      "learning_rate": 3.362555468886093e-05,
      "loss": 0.0627,
      "step": 2862,
      "task_loss": 0.006360730156302452
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.771216879288255,
      "compression_loss": 0.0,
      "distillation_loss": 0.10119855403900146,
      "epoch": 2.72,
      "learning_rate": 3.361555295037019e-05,
      "loss": 0.0967,
      "step": 2863,
      "task_loss": 0.05646238476037979
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7712324013735102,
      "compression_loss": 0.0,
      "distillation_loss": 0.10500767827033997,
      "epoch": 2.72,
      "learning_rate": 3.360554964682276e-05,
      "loss": 0.1036,
      "step": 2864,
      "task_loss": 0.09080617129802704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7712479178772893,
      "compression_loss": 0.0,
      "distillation_loss": 0.025210561230778694,
      "epoch": 2.72,
      "learning_rate": 3.359554478003579e-05,
      "loss": 0.0285,
      "step": 2865,
      "task_loss": 0.0582006499171257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.771263428800596,
      "compression_loss": 0.0,
      "distillation_loss": 0.16816392540931702,
      "epoch": 2.72,
      "learning_rate": 3.358553835182673e-05,
      "loss": 0.1681,
      "step": 2866,
      "task_loss": 0.16722428798675537
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7712789341444338,
      "compression_loss": 0.0,
      "distillation_loss": 0.054264236241579056,
      "epoch": 2.72,
      "learning_rate": 3.357553036401326e-05,
      "loss": 0.0504,
      "step": 2867,
      "task_loss": 0.015533113852143288
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7712944339098066,
      "compression_loss": 0.0,
      "distillation_loss": 0.11492185294628143,
      "epoch": 2.72,
      "learning_rate": 3.356552081841341e-05,
      "loss": 0.1207,
      "step": 2868,
      "task_loss": 0.1728590577840805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.771309928097718,
      "compression_loss": 0.0,
      "distillation_loss": 0.19074279069900513,
      "epoch": 2.72,
      "learning_rate": 3.355550971684545e-05,
      "loss": 0.191,
      "step": 2869,
      "task_loss": 0.19339722394943237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7713254167091717,
      "compression_loss": 0.0,
      "distillation_loss": 0.09511333703994751,
      "epoch": 2.73,
      "learning_rate": 3.3545497061127946e-05,
      "loss": 0.095,
      "step": 2870,
      "task_loss": 0.0938149243593216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7713408997451714,
      "compression_loss": 0.0,
      "distillation_loss": 0.03391757979989052,
      "epoch": 2.73,
      "learning_rate": 3.353548285307975e-05,
      "loss": 0.0316,
      "step": 2871,
      "task_loss": 0.011141767725348473
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7713563772067207,
      "compression_loss": 0.0,
      "distillation_loss": 0.04854762554168701,
      "epoch": 2.73,
      "learning_rate": 3.352546709451998e-05,
      "loss": 0.076,
      "step": 2872,
      "task_loss": 0.3228394389152527
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7713718490948234,
      "compression_loss": 0.0,
      "distillation_loss": 0.07784444838762283,
      "epoch": 2.73,
      "learning_rate": 3.351544978726805e-05,
      "loss": 0.0753,
      "step": 2873,
      "task_loss": 0.05243542417883873
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.771387315410483,
      "compression_loss": 0.0,
      "distillation_loss": 0.050782375037670135,
      "epoch": 2.73,
      "learning_rate": 3.350543093314366e-05,
      "loss": 0.0468,
      "step": 2874,
      "task_loss": 0.01137268915772438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7714027761547034,
      "compression_loss": 0.0,
      "distillation_loss": 0.1053369790315628,
      "epoch": 2.73,
      "learning_rate": 3.349541053396678e-05,
      "loss": 0.109,
      "step": 2875,
      "task_loss": 0.14205503463745117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7714182313284882,
      "compression_loss": 0.0,
      "distillation_loss": 0.12449757009744644,
      "epoch": 2.73,
      "learning_rate": 3.348538859155766e-05,
      "loss": 0.1181,
      "step": 2876,
      "task_loss": 0.060163747519254684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.771433680932841,
      "compression_loss": 0.0,
      "distillation_loss": 0.16609831154346466,
      "epoch": 2.73,
      "learning_rate": 3.347536510773684e-05,
      "loss": 0.1762,
      "step": 2877,
      "task_loss": 0.2674504816532135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7714491249687656,
      "compression_loss": 0.0,
      "distillation_loss": 0.015941990539431572,
      "epoch": 2.73,
      "learning_rate": 3.346534008432513e-05,
      "loss": 0.0148,
      "step": 2878,
      "task_loss": 0.004297057166695595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7714645634372655,
      "compression_loss": 0.0,
      "distillation_loss": 0.06580062955617905,
      "epoch": 2.73,
      "learning_rate": 3.3455313523143615e-05,
      "loss": 0.0644,
      "step": 2879,
      "task_loss": 0.05226878076791763
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7714799963393447,
      "compression_loss": 0.0,
      "distillation_loss": 0.041852548718452454,
      "epoch": 2.74,
      "learning_rate": 3.3445285426013685e-05,
      "loss": 0.0441,
      "step": 2880,
      "task_loss": 0.0639595314860344
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7714954236760067,
      "compression_loss": 0.0,
      "distillation_loss": 0.14709192514419556,
      "epoch": 2.74,
      "learning_rate": 3.343525579475698e-05,
      "loss": 0.1519,
      "step": 2881,
      "task_loss": 0.19560851156711578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.771510845448255,
      "compression_loss": 0.0,
      "distillation_loss": 0.02442905679345131,
      "epoch": 2.74,
      "learning_rate": 3.342522463119543e-05,
      "loss": 0.0226,
      "step": 2882,
      "task_loss": 0.00585402175784111
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7715262616570936,
      "compression_loss": 0.0,
      "distillation_loss": 0.08210425078868866,
      "epoch": 2.74,
      "learning_rate": 3.341519193715127e-05,
      "loss": 0.087,
      "step": 2883,
      "task_loss": 0.13087433576583862
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.771541672303526,
      "compression_loss": 0.0,
      "distillation_loss": 0.021984929218888283,
      "epoch": 2.74,
      "learning_rate": 3.340515771444695e-05,
      "loss": 0.0207,
      "step": 2884,
      "task_loss": 0.008725512772798538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.771557077388556,
      "compression_loss": 0.0,
      "distillation_loss": 0.1168016642332077,
      "epoch": 2.74,
      "learning_rate": 3.3395121964905265e-05,
      "loss": 0.1205,
      "step": 2885,
      "task_loss": 0.15377360582351685
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.771572476913187,
      "compression_loss": 0.0,
      "distillation_loss": 0.16023914515972137,
      "epoch": 2.74,
      "learning_rate": 3.338508469034922e-05,
      "loss": 0.1587,
      "step": 2886,
      "task_loss": 0.14454258978366852
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7715878708784231,
      "compression_loss": 0.0,
      "distillation_loss": 0.05907361954450607,
      "epoch": 2.74,
      "learning_rate": 3.3375045892602176e-05,
      "loss": 0.0606,
      "step": 2887,
      "task_loss": 0.07458890229463577
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7716032592852676,
      "compression_loss": 0.0,
      "distillation_loss": 0.0668545514345169,
      "epoch": 2.74,
      "learning_rate": 3.3365005573487706e-05,
      "loss": 0.0705,
      "step": 2888,
      "task_loss": 0.10356783866882324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7716186421347244,
      "compression_loss": 0.0,
      "distillation_loss": 0.052939146757125854,
      "epoch": 2.74,
      "learning_rate": 3.335496373482969e-05,
      "loss": 0.0487,
      "step": 2889,
      "task_loss": 0.010425196960568428
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7716340194277972,
      "compression_loss": 0.0,
      "distillation_loss": 0.022259045392274857,
      "epoch": 2.74,
      "learning_rate": 3.334492037845227e-05,
      "loss": 0.0272,
      "step": 2890,
      "task_loss": 0.07185941934585571
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7716493911654896,
      "compression_loss": 0.0,
      "distillation_loss": 0.05123762786388397,
      "epoch": 2.75,
      "learning_rate": 3.333487550617987e-05,
      "loss": 0.0698,
      "step": 2891,
      "task_loss": 0.2367219626903534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7716647573488052,
      "compression_loss": 0.0,
      "distillation_loss": 0.18512699007987976,
      "epoch": 2.75,
      "learning_rate": 3.332482911983721e-05,
      "loss": 0.1761,
      "step": 2892,
      "task_loss": 0.09492430090904236
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7716801179787479,
      "compression_loss": 0.0,
      "distillation_loss": 0.18022358417510986,
      "epoch": 2.75,
      "learning_rate": 3.331478122124924e-05,
      "loss": 0.1795,
      "step": 2893,
      "task_loss": 0.1729545146226883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7716954730563212,
      "compression_loss": 0.0,
      "distillation_loss": 0.020589305087924004,
      "epoch": 2.75,
      "learning_rate": 3.330473181224121e-05,
      "loss": 0.0191,
      "step": 2894,
      "task_loss": 0.005505530163645744
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7717108225825289,
      "compression_loss": 0.0,
      "distillation_loss": 0.030846502631902695,
      "epoch": 2.75,
      "learning_rate": 3.3294680894638655e-05,
      "loss": 0.0285,
      "step": 2895,
      "task_loss": 0.0070547014474868774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7717261665583746,
      "compression_loss": 0.0,
      "distillation_loss": 0.020820213481783867,
      "epoch": 2.75,
      "learning_rate": 3.328462847026736e-05,
      "loss": 0.0287,
      "step": 2896,
      "task_loss": 0.09968282282352448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7717415049848619,
      "compression_loss": 0.0,
      "distillation_loss": 0.09029317647218704,
      "epoch": 2.75,
      "learning_rate": 3.327457454095342e-05,
      "loss": 0.0924,
      "step": 2897,
      "task_loss": 0.11134281009435654
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7717568378629946,
      "compression_loss": 0.0,
      "distillation_loss": 0.06361924111843109,
      "epoch": 2.75,
      "learning_rate": 3.3264519108523154e-05,
      "loss": 0.0595,
      "step": 2898,
      "task_loss": 0.022628160193562508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7717721651937766,
      "compression_loss": 0.0,
      "distillation_loss": 0.04663745313882828,
      "epoch": 2.75,
      "learning_rate": 3.3254462174803186e-05,
      "loss": 0.048,
      "step": 2899,
      "task_loss": 0.060523584485054016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7717874869782111,
      "compression_loss": 0.0,
      "distillation_loss": 0.23908516764640808,
      "epoch": 2.75,
      "learning_rate": 3.324440374162041e-05,
      "loss": 0.2375,
      "step": 2900,
      "task_loss": 0.22366894781589508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7718028032173022,
      "compression_loss": 0.0,
      "distillation_loss": 0.07372735440731049,
      "epoch": 2.75,
      "learning_rate": 3.323434381080199e-05,
      "loss": 0.0802,
      "step": 2901,
      "task_loss": 0.13813573122024536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7718181139120533,
      "compression_loss": 0.0,
      "distillation_loss": 0.04444555193185806,
      "epoch": 2.76,
      "learning_rate": 3.322428238417537e-05,
      "loss": 0.0548,
      "step": 2902,
      "task_loss": 0.1481703817844391
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7718334190634683,
      "compression_loss": 0.0,
      "distillation_loss": 0.10492336750030518,
      "epoch": 2.76,
      "learning_rate": 3.321421946356823e-05,
      "loss": 0.0998,
      "step": 2903,
      "task_loss": 0.05350606143474579
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7718487186725508,
      "compression_loss": 0.0,
      "distillation_loss": 0.043171461671590805,
      "epoch": 2.76,
      "learning_rate": 3.320415505080858e-05,
      "loss": 0.0399,
      "step": 2904,
      "task_loss": 0.010258587077260017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7718640127403044,
      "compression_loss": 0.0,
      "distillation_loss": 0.09338559955358505,
      "epoch": 2.76,
      "learning_rate": 3.3194089147724644e-05,
      "loss": 0.0905,
      "step": 2905,
      "task_loss": 0.0642678365111351
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7718793012677327,
      "compression_loss": 0.0,
      "distillation_loss": 0.07148714363574982,
      "epoch": 2.76,
      "learning_rate": 3.3184021756144954e-05,
      "loss": 0.0692,
      "step": 2906,
      "task_loss": 0.048625778406858444
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7718945842558398,
      "compression_loss": 0.0,
      "distillation_loss": 0.053975023329257965,
      "epoch": 2.76,
      "learning_rate": 3.317395287789829e-05,
      "loss": 0.0522,
      "step": 2907,
      "task_loss": 0.03605438768863678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7719098617056289,
      "compression_loss": 0.0,
      "distillation_loss": 0.21236878633499146,
      "epoch": 2.76,
      "learning_rate": 3.316388251481373e-05,
      "loss": 0.2012,
      "step": 2908,
      "task_loss": 0.10068190097808838
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.771925133618104,
      "compression_loss": 0.0,
      "distillation_loss": 0.08693385124206543,
      "epoch": 2.76,
      "learning_rate": 3.3153810668720594e-05,
      "loss": 0.0805,
      "step": 2909,
      "task_loss": 0.022921759635210037
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7719403999942687,
      "compression_loss": 0.0,
      "distillation_loss": 0.07356773316860199,
      "epoch": 2.76,
      "learning_rate": 3.3143737341448475e-05,
      "loss": 0.0745,
      "step": 2910,
      "task_loss": 0.08280356228351593
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7719556608351267,
      "compression_loss": 0.0,
      "distillation_loss": 0.03629321604967117,
      "epoch": 2.76,
      "learning_rate": 3.3133662534827255e-05,
      "loss": 0.034,
      "step": 2911,
      "task_loss": 0.01335129514336586
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7719709161416815,
      "compression_loss": 0.0,
      "distillation_loss": 0.08336710929870605,
      "epoch": 2.77,
      "learning_rate": 3.3123586250687055e-05,
      "loss": 0.078,
      "step": 2912,
      "task_loss": 0.02927432768046856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.771986165914937,
      "compression_loss": 0.0,
      "distillation_loss": 0.07923730462789536,
      "epoch": 2.77,
      "learning_rate": 3.311350849085829e-05,
      "loss": 0.0754,
      "step": 2913,
      "task_loss": 0.04064404219388962
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7720014101558967,
      "compression_loss": 0.0,
      "distillation_loss": 0.1845068484544754,
      "epoch": 2.77,
      "learning_rate": 3.3103429257171635e-05,
      "loss": 0.1826,
      "step": 2914,
      "task_loss": 0.16558828949928284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7720166488655645,
      "compression_loss": 0.0,
      "distillation_loss": 0.03765151649713516,
      "epoch": 2.77,
      "learning_rate": 3.309334855145803e-05,
      "loss": 0.043,
      "step": 2915,
      "task_loss": 0.09107490628957748
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7720318820449439,
      "compression_loss": 0.0,
      "distillation_loss": 0.1513894945383072,
      "epoch": 2.77,
      "learning_rate": 3.3083266375548675e-05,
      "loss": 0.1645,
      "step": 2916,
      "task_loss": 0.2826227843761444
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7720471096950388,
      "compression_loss": 0.0,
      "distillation_loss": 0.10186254978179932,
      "epoch": 2.77,
      "learning_rate": 3.3073182731275064e-05,
      "loss": 0.098,
      "step": 2917,
      "task_loss": 0.06284405291080475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7720623318168525,
      "compression_loss": 0.0,
      "distillation_loss": 0.18332624435424805,
      "epoch": 2.77,
      "learning_rate": 3.306309762046892e-05,
      "loss": 0.1762,
      "step": 2918,
      "task_loss": 0.1121426373720169
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7720775484113891,
      "compression_loss": 0.0,
      "distillation_loss": 0.0718914344906807,
      "epoch": 2.77,
      "learning_rate": 3.305301104496227e-05,
      "loss": 0.0692,
      "step": 2919,
      "task_loss": 0.04449700936675072
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.772092759479652,
      "compression_loss": 0.0,
      "distillation_loss": 0.19091036915779114,
      "epoch": 2.77,
      "learning_rate": 3.3042923006587366e-05,
      "loss": 0.1988,
      "step": 2920,
      "task_loss": 0.26946884393692017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.772107965022645,
      "compression_loss": 0.0,
      "distillation_loss": 0.05884380638599396,
      "epoch": 2.77,
      "learning_rate": 3.303283350717678e-05,
      "loss": 0.0645,
      "step": 2921,
      "task_loss": 0.11569136381149292
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7721231650413717,
      "compression_loss": 0.0,
      "distillation_loss": 0.06322423368692398,
      "epoch": 2.77,
      "learning_rate": 3.302274254856329e-05,
      "loss": 0.06,
      "step": 2922,
      "task_loss": 0.031089693307876587
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.772138359536836,
      "compression_loss": 0.0,
      "distillation_loss": 0.13263660669326782,
      "epoch": 2.78,
      "learning_rate": 3.301265013257998e-05,
      "loss": 0.1316,
      "step": 2923,
      "task_loss": 0.12230206280946732
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7721535485100413,
      "compression_loss": 0.0,
      "distillation_loss": 0.1299370527267456,
      "epoch": 2.78,
      "learning_rate": 3.300255626106019e-05,
      "loss": 0.1243,
      "step": 2924,
      "task_loss": 0.07319293916225433
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7721687319619914,
      "compression_loss": 0.0,
      "distillation_loss": 0.024539334699511528,
      "epoch": 2.78,
      "learning_rate": 3.2992460935837505e-05,
      "loss": 0.0226,
      "step": 2925,
      "task_loss": 0.005514673888683319
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.77218390989369,
      "compression_loss": 0.0,
      "distillation_loss": 0.14706680178642273,
      "epoch": 2.78,
      "learning_rate": 3.2982364158745805e-05,
      "loss": 0.1399,
      "step": 2926,
      "task_loss": 0.07497966289520264
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7721990823061408,
      "compression_loss": 0.0,
      "distillation_loss": 0.07125681638717651,
      "epoch": 2.78,
      "learning_rate": 3.297226593161921e-05,
      "loss": 0.0669,
      "step": 2927,
      "task_loss": 0.027698028832674026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7722142492003474,
      "compression_loss": 0.0,
      "distillation_loss": 0.25317683815956116,
      "epoch": 2.78,
      "learning_rate": 3.2962166256292113e-05,
      "loss": 0.2522,
      "step": 2928,
      "task_loss": 0.24299074709415436
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7722294105773136,
      "compression_loss": 0.0,
      "distillation_loss": 0.10056150704622269,
      "epoch": 2.78,
      "learning_rate": 3.295206513459917e-05,
      "loss": 0.0971,
      "step": 2929,
      "task_loss": 0.06555324792861938
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.772244566438043,
      "compression_loss": 0.0,
      "distillation_loss": 0.10626856982707977,
      "epoch": 2.78,
      "learning_rate": 3.2941962568375296e-05,
      "loss": 0.1043,
      "step": 2930,
      "task_loss": 0.08672097325325012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7722597167835393,
      "compression_loss": 0.0,
      "distillation_loss": 0.02176981046795845,
      "epoch": 2.78,
      "learning_rate": 3.2931858559455674e-05,
      "loss": 0.043,
      "step": 2931,
      "task_loss": 0.23452335596084595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7722748616148062,
      "compression_loss": 0.0,
      "distillation_loss": 0.26498720049858093,
      "epoch": 2.78,
      "learning_rate": 3.292175310967575e-05,
      "loss": 0.2554,
      "step": 2932,
      "task_loss": 0.16884511709213257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7722900009328473,
      "compression_loss": 0.0,
      "distillation_loss": 0.020644748583436012,
      "epoch": 2.79,
      "learning_rate": 3.291164622087122e-05,
      "loss": 0.0191,
      "step": 2933,
      "task_loss": 0.005039013922214508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7723051347386664,
      "compression_loss": 0.0,
      "distillation_loss": 0.02696690335869789,
      "epoch": 2.79,
      "learning_rate": 3.290153789487804e-05,
      "loss": 0.0264,
      "step": 2934,
      "task_loss": 0.02141895517706871
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7723202630332671,
      "compression_loss": 0.0,
      "distillation_loss": 0.056153517216444016,
      "epoch": 2.79,
      "learning_rate": 3.289142813353246e-05,
      "loss": 0.0626,
      "step": 2935,
      "task_loss": 0.12072312086820602
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7723353858176532,
      "compression_loss": 0.0,
      "distillation_loss": 0.09314046055078506,
      "epoch": 2.79,
      "learning_rate": 3.2881316938670945e-05,
      "loss": 0.1048,
      "step": 2936,
      "task_loss": 0.20927327871322632
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7723505030928282,
      "compression_loss": 0.0,
      "distillation_loss": 0.02801085263490677,
      "epoch": 2.79,
      "learning_rate": 3.2871204312130254e-05,
      "loss": 0.0387,
      "step": 2937,
      "task_loss": 0.13507309556007385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7723656148597958,
      "compression_loss": 0.0,
      "distillation_loss": 0.0866575613617897,
      "epoch": 2.79,
      "learning_rate": 3.28610902557474e-05,
      "loss": 0.0907,
      "step": 2938,
      "task_loss": 0.12749417126178741
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7723807211195599,
      "compression_loss": 0.0,
      "distillation_loss": 0.11423015594482422,
      "epoch": 2.79,
      "learning_rate": 3.285097477135963e-05,
      "loss": 0.1059,
      "step": 2939,
      "task_loss": 0.030799927189946175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7723958218731239,
      "compression_loss": 0.0,
      "distillation_loss": 0.10484655201435089,
      "epoch": 2.79,
      "learning_rate": 3.284085786080449e-05,
      "loss": 0.1022,
      "step": 2940,
      "task_loss": 0.07836197316646576
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7724109171214917,
      "compression_loss": 0.0,
      "distillation_loss": 0.03331822156906128,
      "epoch": 2.79,
      "learning_rate": 3.283073952591975e-05,
      "loss": 0.0453,
      "step": 2941,
      "task_loss": 0.1528467983007431
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.772426006865667,
      "compression_loss": 0.0,
      "distillation_loss": 0.15390589833259583,
      "epoch": 2.79,
      "learning_rate": 3.2820619768543473e-05,
      "loss": 0.1486,
      "step": 2942,
      "task_loss": 0.10106901824474335
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7724410911066533,
      "compression_loss": 0.0,
      "distillation_loss": 0.02205563150346279,
      "epoch": 2.79,
      "learning_rate": 3.281049859051394e-05,
      "loss": 0.0202,
      "step": 2943,
      "task_loss": 0.003351377323269844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7724561698454543,
      "compression_loss": 0.0,
      "distillation_loss": 0.04081364721059799,
      "epoch": 2.8,
      "learning_rate": 3.280037599366972e-05,
      "loss": 0.039,
      "step": 2944,
      "task_loss": 0.022568656131625175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7724712430830738,
      "compression_loss": 0.0,
      "distillation_loss": 0.10027671605348587,
      "epoch": 2.8,
      "learning_rate": 3.2790251979849654e-05,
      "loss": 0.0971,
      "step": 2945,
      "task_loss": 0.06809482723474503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7724863108205153,
      "compression_loss": 0.0,
      "distillation_loss": 0.11592331528663635,
      "epoch": 2.8,
      "learning_rate": 3.278012655089277e-05,
      "loss": 0.1258,
      "step": 2946,
      "task_loss": 0.21512287855148315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7725013730587829,
      "compression_loss": 0.0,
      "distillation_loss": 0.11402192711830139,
      "epoch": 2.8,
      "learning_rate": 3.276999970863845e-05,
      "loss": 0.1178,
      "step": 2947,
      "task_loss": 0.1516398787498474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7725164297988798,
      "compression_loss": 0.0,
      "distillation_loss": 0.039229217916727066,
      "epoch": 2.8,
      "learning_rate": 3.275987145492625e-05,
      "loss": 0.0364,
      "step": 2948,
      "task_loss": 0.010563917458057404
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7725314810418099,
      "compression_loss": 0.0,
      "distillation_loss": 0.179963618516922,
      "epoch": 2.8,
      "learning_rate": 3.274974179159603e-05,
      "loss": 0.181,
      "step": 2949,
      "task_loss": 0.1902376413345337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7725465267885768,
      "compression_loss": 0.0,
      "distillation_loss": 0.08978709578514099,
      "epoch": 2.8,
      "learning_rate": 3.27396107204879e-05,
      "loss": 0.0879,
      "step": 2950,
      "task_loss": 0.07117792963981628
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7725615670401843,
      "compression_loss": 0.0,
      "distillation_loss": 0.15075881779193878,
      "epoch": 2.8,
      "learning_rate": 3.2729478243442194e-05,
      "loss": 0.1489,
      "step": 2951,
      "task_loss": 0.13251779973506927
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7725766017976361,
      "compression_loss": 0.0,
      "distillation_loss": 0.07806205749511719,
      "epoch": 2.8,
      "learning_rate": 3.271934436229955e-05,
      "loss": 0.0817,
      "step": 2952,
      "task_loss": 0.11416235566139221
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7725916310619357,
      "compression_loss": 0.0,
      "distillation_loss": 0.23699919879436493,
      "epoch": 2.8,
      "learning_rate": 3.270920907890082e-05,
      "loss": 0.2343,
      "step": 2953,
      "task_loss": 0.21003299951553345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.772606654834087,
      "compression_loss": 0.0,
      "distillation_loss": 0.09724032133817673,
      "epoch": 2.81,
      "learning_rate": 3.269907239508714e-05,
      "loss": 0.0913,
      "step": 2954,
      "task_loss": 0.037794340401887894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7726216731150934,
      "compression_loss": 0.0,
      "distillation_loss": 0.11641349643468857,
      "epoch": 2.81,
      "learning_rate": 3.268893431269987e-05,
      "loss": 0.1332,
      "step": 2955,
      "task_loss": 0.2846828103065491
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7726366859059589,
      "compression_loss": 0.0,
      "distillation_loss": 0.13409125804901123,
      "epoch": 2.81,
      "learning_rate": 3.2678794833580654e-05,
      "loss": 0.1328,
      "step": 2956,
      "task_loss": 0.1211758628487587
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7726516932076869,
      "compression_loss": 0.0,
      "distillation_loss": 0.1196846067905426,
      "epoch": 2.81,
      "learning_rate": 3.2668653959571384e-05,
      "loss": 0.1139,
      "step": 2957,
      "task_loss": 0.06228470057249069
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7726666950212814,
      "compression_loss": 0.0,
      "distillation_loss": 0.12382517009973526,
      "epoch": 2.81,
      "learning_rate": 3.2658511692514184e-05,
      "loss": 0.1171,
      "step": 2958,
      "task_loss": 0.056750208139419556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7726816913477457,
      "compression_loss": 0.0,
      "distillation_loss": 0.07181763648986816,
      "epoch": 2.81,
      "learning_rate": 3.2648368034251454e-05,
      "loss": 0.073,
      "step": 2959,
      "task_loss": 0.0838204026222229
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7726966821880837,
      "compression_loss": 0.0,
      "distillation_loss": 0.19776178896427155,
      "epoch": 2.81,
      "learning_rate": 3.263822298662583e-05,
      "loss": 0.1927,
      "step": 2960,
      "task_loss": 0.14683988690376282
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7727116675432992,
      "compression_loss": 0.0,
      "distillation_loss": 0.1264166533946991,
      "epoch": 2.81,
      "learning_rate": 3.2628076551480216e-05,
      "loss": 0.1285,
      "step": 2961,
      "task_loss": 0.14740031957626343
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7727266474143957,
      "compression_loss": 0.0,
      "distillation_loss": 0.18037302792072296,
      "epoch": 2.81,
      "learning_rate": 3.2617928730657764e-05,
      "loss": 0.189,
      "step": 2962,
      "task_loss": 0.2670821249485016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7727416218023769,
      "compression_loss": 0.0,
      "distillation_loss": 0.02856997214257717,
      "epoch": 2.81,
      "learning_rate": 3.260777952600186e-05,
      "loss": 0.0308,
      "step": 2963,
      "task_loss": 0.051173169165849686
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7727565907082465,
      "compression_loss": 0.0,
      "distillation_loss": 0.20055626332759857,
      "epoch": 2.81,
      "learning_rate": 3.2597628939356175e-05,
      "loss": 0.1923,
      "step": 2964,
      "task_loss": 0.11836535483598709
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7727715541330082,
      "compression_loss": 0.0,
      "distillation_loss": 0.10277973115444183,
      "epoch": 2.82,
      "learning_rate": 3.25874769725646e-05,
      "loss": 0.1018,
      "step": 2965,
      "task_loss": 0.09281755983829498
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7727865120776657,
      "compression_loss": 0.0,
      "distillation_loss": 0.07278560847043991,
      "epoch": 2.82,
      "learning_rate": 3.257732362747129e-05,
      "loss": 0.0731,
      "step": 2966,
      "task_loss": 0.07589083909988403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7728014645432226,
      "compression_loss": 0.0,
      "distillation_loss": 0.05969984084367752,
      "epoch": 2.82,
      "learning_rate": 3.256716890592065e-05,
      "loss": 0.0624,
      "step": 2967,
      "task_loss": 0.08681820333003998
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7728164115306826,
      "compression_loss": 0.0,
      "distillation_loss": 0.1937519758939743,
      "epoch": 2.82,
      "learning_rate": 3.255701280975733e-05,
      "loss": 0.1939,
      "step": 2968,
      "task_loss": 0.19502577185630798
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7728313530410496,
      "compression_loss": 0.0,
      "distillation_loss": 0.14739632606506348,
      "epoch": 2.82,
      "learning_rate": 3.2546855340826246e-05,
      "loss": 0.1487,
      "step": 2969,
      "task_loss": 0.1602475941181183
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7728462890753269,
      "compression_loss": 0.0,
      "distillation_loss": 0.16489112377166748,
      "epoch": 2.82,
      "learning_rate": 3.253669650097254e-05,
      "loss": 0.1558,
      "step": 2970,
      "task_loss": 0.07406702637672424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7728612196345185,
      "compression_loss": 0.0,
      "distillation_loss": 0.10662008821964264,
      "epoch": 2.82,
      "learning_rate": 3.2526536292041625e-05,
      "loss": 0.1111,
      "step": 2971,
      "task_loss": 0.1511189192533493
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.772876144719628,
      "compression_loss": 0.0,
      "distillation_loss": 0.0431804321706295,
      "epoch": 2.82,
      "learning_rate": 3.2516374715879126e-05,
      "loss": 0.0418,
      "step": 2972,
      "task_loss": 0.02895108424127102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7728910643316589,
      "compression_loss": 0.0,
      "distillation_loss": 0.1767425686120987,
      "epoch": 2.82,
      "learning_rate": 3.250621177433097e-05,
      "loss": 0.1788,
      "step": 2973,
      "task_loss": 0.197010338306427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.772905978471615,
      "compression_loss": 0.0,
      "distillation_loss": 0.12274418771266937,
      "epoch": 2.82,
      "learning_rate": 3.249604746924331e-05,
      "loss": 0.1189,
      "step": 2974,
      "task_loss": 0.08417470753192902
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7729208871405002,
      "compression_loss": 0.0,
      "distillation_loss": 0.023153632879257202,
      "epoch": 2.83,
      "learning_rate": 3.248588180246251e-05,
      "loss": 0.0321,
      "step": 2975,
      "task_loss": 0.11221934109926224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7729357903393179,
      "compression_loss": 0.0,
      "distillation_loss": 0.24052870273590088,
      "epoch": 2.83,
      "learning_rate": 3.247571477583523e-05,
      "loss": 0.248,
      "step": 2976,
      "task_loss": 0.31512880325317383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7729506880690719,
      "compression_loss": 0.0,
      "distillation_loss": 0.11706255376338959,
      "epoch": 2.83,
      "learning_rate": 3.2465546391208355e-05,
      "loss": 0.1414,
      "step": 2977,
      "task_loss": 0.36023640632629395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7729655803307658,
      "compression_loss": 0.0,
      "distillation_loss": 0.24018515646457672,
      "epoch": 2.83,
      "learning_rate": 3.245537665042903e-05,
      "loss": 0.2329,
      "step": 2978,
      "task_loss": 0.1674792766571045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7729804671254034,
      "compression_loss": 0.0,
      "distillation_loss": 0.12150129675865173,
      "epoch": 2.83,
      "learning_rate": 3.244520555534463e-05,
      "loss": 0.1159,
      "step": 2979,
      "task_loss": 0.06535904854536057
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7729953484539883,
      "compression_loss": 0.0,
      "distillation_loss": 0.09363529086112976,
      "epoch": 2.83,
      "learning_rate": 3.243503310780278e-05,
      "loss": 0.0944,
      "step": 2980,
      "task_loss": 0.10123631358146667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7730102243175242,
      "compression_loss": 0.0,
      "distillation_loss": 0.060428209602832794,
      "epoch": 2.83,
      "learning_rate": 3.242485930965136e-05,
      "loss": 0.0578,
      "step": 2981,
      "task_loss": 0.03405189514160156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7730250947170147,
      "compression_loss": 0.0,
      "distillation_loss": 0.03489404916763306,
      "epoch": 2.83,
      "learning_rate": 3.241468416273849e-05,
      "loss": 0.0514,
      "step": 2982,
      "task_loss": 0.2001369595527649
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7730399596534637,
      "compression_loss": 0.0,
      "distillation_loss": 0.026621265336871147,
      "epoch": 2.83,
      "learning_rate": 3.2404507668912534e-05,
      "loss": 0.0257,
      "step": 2983,
      "task_loss": 0.017897440120577812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7730548191278748,
      "compression_loss": 0.0,
      "distillation_loss": 0.06251438707113266,
      "epoch": 2.83,
      "learning_rate": 3.2394329830022095e-05,
      "loss": 0.0658,
      "step": 2984,
      "task_loss": 0.0951443687081337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7730696731412514,
      "compression_loss": 0.0,
      "distillation_loss": 0.06263174116611481,
      "epoch": 2.83,
      "learning_rate": 3.238415064791603e-05,
      "loss": 0.0604,
      "step": 2985,
      "task_loss": 0.03981752693653107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7730845216945976,
      "compression_loss": 0.0,
      "distillation_loss": 0.04605867341160774,
      "epoch": 2.84,
      "learning_rate": 3.237397012444344e-05,
      "loss": 0.056,
      "step": 2986,
      "task_loss": 0.145818829536438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7730993647889168,
      "compression_loss": 0.0,
      "distillation_loss": 0.08241648226976395,
      "epoch": 2.84,
      "learning_rate": 3.2363788261453664e-05,
      "loss": 0.0888,
      "step": 2987,
      "task_loss": 0.14630846679210663
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7731142024252129,
      "compression_loss": 0.0,
      "distillation_loss": 0.027548307552933693,
      "epoch": 2.84,
      "learning_rate": 3.2353605060796286e-05,
      "loss": 0.0253,
      "step": 2988,
      "task_loss": 0.004967987537384033
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7731290346044893,
      "compression_loss": 0.0,
      "distillation_loss": 0.17513786256313324,
      "epoch": 2.84,
      "learning_rate": 3.2343420524321134e-05,
      "loss": 0.1683,
      "step": 2989,
      "task_loss": 0.1069210097193718
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.77314386132775,
      "compression_loss": 0.0,
      "distillation_loss": 0.0309540294110775,
      "epoch": 2.84,
      "learning_rate": 3.2333234653878275e-05,
      "loss": 0.033,
      "step": 2990,
      "task_loss": 0.051752254366874695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7731586825959984,
      "compression_loss": 0.0,
      "distillation_loss": 0.07692794501781464,
      "epoch": 2.84,
      "learning_rate": 3.2323047451318023e-05,
      "loss": 0.0828,
      "step": 2991,
      "task_loss": 0.1354198455810547
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7731734984102384,
      "compression_loss": 0.0,
      "distillation_loss": 0.23258818686008453,
      "epoch": 2.84,
      "learning_rate": 3.2312858918490936e-05,
      "loss": 0.2235,
      "step": 2992,
      "task_loss": 0.1417001634836197
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7731883087714736,
      "compression_loss": 0.0,
      "distillation_loss": 0.07296988368034363,
      "epoch": 2.84,
      "learning_rate": 3.2302669057247806e-05,
      "loss": 0.0679,
      "step": 2993,
      "task_loss": 0.02189962938427925
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7732031136807076,
      "compression_loss": 0.0,
      "distillation_loss": 0.13466548919677734,
      "epoch": 2.84,
      "learning_rate": 3.2292477869439666e-05,
      "loss": 0.1294,
      "step": 2994,
      "task_loss": 0.08172232657670975
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7732179131389442,
      "compression_loss": 0.0,
      "distillation_loss": 0.1007700264453888,
      "epoch": 2.84,
      "learning_rate": 3.228228535691781e-05,
      "loss": 0.0913,
      "step": 2995,
      "task_loss": 0.0059202127158641815
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.773232707147187,
      "compression_loss": 0.0,
      "distillation_loss": 0.13270485401153564,
      "epoch": 2.85,
      "learning_rate": 3.2272091521533745e-05,
      "loss": 0.1522,
      "step": 2996,
      "task_loss": 0.3279738426208496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7732474957064397,
      "compression_loss": 0.0,
      "distillation_loss": 0.09864410758018494,
      "epoch": 2.85,
      "learning_rate": 3.226189636513923e-05,
      "loss": 0.0946,
      "step": 2997,
      "task_loss": 0.05842628702521324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.773262278817706,
      "compression_loss": 0.0,
      "distillation_loss": 0.14679734408855438,
      "epoch": 2.85,
      "learning_rate": 3.225169988958627e-05,
      "loss": 0.1396,
      "step": 2998,
      "task_loss": 0.07466141879558563
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7732770564819896,
      "compression_loss": 0.0,
      "distillation_loss": 0.09835858643054962,
      "epoch": 2.85,
      "learning_rate": 3.2241502096727095e-05,
      "loss": 0.0922,
      "step": 2999,
      "task_loss": 0.03713885694742203
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7732918287002941,
      "compression_loss": 0.0,
      "distillation_loss": 0.08745696395635605,
      "epoch": 2.85,
      "learning_rate": 3.2231302988414194e-05,
      "loss": 0.0818,
      "step": 3000,
      "task_loss": 0.030404910445213318
    },
    {
      "epoch": 2.85,
      "eval_accuracy": 0.8899082568807339,
      "eval_loss": 0.41417789459228516,
      "eval_runtime": 17.9806,
      "eval_samples_per_second": 48.497,
      "eval_steps_per_second": 6.062,
      "step": 3000
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7733065954736233,
      "compression_loss": 0.0,
      "distillation_loss": 0.16211673617362976,
      "epoch": 2.85,
      "learning_rate": 3.222110256650028e-05,
      "loss": 0.1516,
      "step": 3001,
      "task_loss": 0.05707669258117676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7733213568029809,
      "compression_loss": 0.0,
      "distillation_loss": 0.10503898561000824,
      "epoch": 2.85,
      "learning_rate": 3.2210900832838295e-05,
      "loss": 0.1082,
      "step": 3002,
      "task_loss": 0.13650718331336975
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7733361126893704,
      "compression_loss": 0.0,
      "distillation_loss": 0.037988051772117615,
      "epoch": 2.85,
      "learning_rate": 3.220069778928146e-05,
      "loss": 0.0421,
      "step": 3003,
      "task_loss": 0.0787007063627243
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7733508631337956,
      "compression_loss": 0.0,
      "distillation_loss": 0.10283227264881134,
      "epoch": 2.85,
      "learning_rate": 3.2190493437683185e-05,
      "loss": 0.1037,
      "step": 3004,
      "task_loss": 0.11105208098888397
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7733656081372603,
      "compression_loss": 0.0,
      "distillation_loss": 0.1534404307603836,
      "epoch": 2.85,
      "learning_rate": 3.2180287779897155e-05,
      "loss": 0.1409,
      "step": 3005,
      "task_loss": 0.028170330449938774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7733803477007679,
      "compression_loss": 0.0,
      "distillation_loss": 0.03912707045674324,
      "epoch": 2.85,
      "learning_rate": 3.217008081777726e-05,
      "loss": 0.0443,
      "step": 3006,
      "task_loss": 0.09075871109962463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7733950818253224,
      "compression_loss": 0.0,
      "distillation_loss": 0.08084475994110107,
      "epoch": 2.86,
      "learning_rate": 3.2159872553177655e-05,
      "loss": 0.0756,
      "step": 3007,
      "task_loss": 0.028212103992700577
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7734098105119271,
      "compression_loss": 0.0,
      "distillation_loss": 0.09292879700660706,
      "epoch": 2.86,
      "learning_rate": 3.2149662987952725e-05,
      "loss": 0.0965,
      "step": 3008,
      "task_loss": 0.12906073033809662
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7734245337615862,
      "compression_loss": 0.0,
      "distillation_loss": 0.04876542091369629,
      "epoch": 2.86,
      "learning_rate": 3.213945212395707e-05,
      "loss": 0.0478,
      "step": 3009,
      "task_loss": 0.039374131709337234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7734392515753028,
      "compression_loss": 0.0,
      "distillation_loss": 0.061346620321273804,
      "epoch": 2.86,
      "learning_rate": 3.212923996304556e-05,
      "loss": 0.0648,
      "step": 3010,
      "task_loss": 0.09587064385414124
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7734539639540811,
      "compression_loss": 0.0,
      "distillation_loss": 0.18387816846370697,
      "epoch": 2.86,
      "learning_rate": 3.211902650707327e-05,
      "loss": 0.1763,
      "step": 3011,
      "task_loss": 0.10823600739240646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7734686708989243,
      "compression_loss": 0.0,
      "distillation_loss": 0.08546897023916245,
      "epoch": 2.86,
      "learning_rate": 3.210881175789553e-05,
      "loss": 0.0837,
      "step": 3012,
      "task_loss": 0.06747668981552124
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7734833724108365,
      "compression_loss": 0.0,
      "distillation_loss": 0.26206955313682556,
      "epoch": 2.86,
      "learning_rate": 3.209859571736791e-05,
      "loss": 0.2582,
      "step": 3013,
      "task_loss": 0.2230200469493866
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7734980684908211,
      "compression_loss": 0.0,
      "distillation_loss": 0.02551981247961521,
      "epoch": 2.86,
      "learning_rate": 3.208837838734618e-05,
      "loss": 0.0314,
      "step": 3014,
      "task_loss": 0.08437542617321014
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.773512759139882,
      "compression_loss": 0.0,
      "distillation_loss": 0.11942901462316513,
      "epoch": 2.86,
      "learning_rate": 3.207815976968638e-05,
      "loss": 0.1173,
      "step": 3015,
      "task_loss": 0.0981997698545456
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7735274443590228,
      "compression_loss": 0.0,
      "distillation_loss": 0.03865540400147438,
      "epoch": 2.86,
      "learning_rate": 3.2067939866244764e-05,
      "loss": 0.0452,
      "step": 3016,
      "task_loss": 0.10382010787725449
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7735421241492471,
      "compression_loss": 0.0,
      "distillation_loss": 0.1100415512919426,
      "epoch": 2.87,
      "learning_rate": 3.205771867887784e-05,
      "loss": 0.113,
      "step": 3017,
      "task_loss": 0.13981683552265167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7735567985115586,
      "compression_loss": 0.0,
      "distillation_loss": 0.19163449108600616,
      "epoch": 2.87,
      "learning_rate": 3.204749620944232e-05,
      "loss": 0.1836,
      "step": 3018,
      "task_loss": 0.11117150634527206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.773571467446961,
      "compression_loss": 0.0,
      "distillation_loss": 0.042591970413923264,
      "epoch": 2.87,
      "learning_rate": 3.203727245979518e-05,
      "loss": 0.0501,
      "step": 3019,
      "task_loss": 0.11770500242710114
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7735861309564581,
      "compression_loss": 0.0,
      "distillation_loss": 0.17520135641098022,
      "epoch": 2.87,
      "learning_rate": 3.20270474317936e-05,
      "loss": 0.173,
      "step": 3020,
      "task_loss": 0.15289539098739624
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7736007890410534,
      "compression_loss": 0.0,
      "distillation_loss": 0.058556366711854935,
      "epoch": 2.87,
      "learning_rate": 3.201682112729502e-05,
      "loss": 0.0701,
      "step": 3021,
      "task_loss": 0.17429102957248688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7736154417017508,
      "compression_loss": 0.0,
      "distillation_loss": 0.12820349633693695,
      "epoch": 2.87,
      "learning_rate": 3.2006593548157074e-05,
      "loss": 0.1274,
      "step": 3022,
      "task_loss": 0.11984600871801376
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7736300889395538,
      "compression_loss": 0.0,
      "distillation_loss": 0.12782520055770874,
      "epoch": 2.87,
      "learning_rate": 3.1996364696237676e-05,
      "loss": 0.1207,
      "step": 3023,
      "task_loss": 0.05618831515312195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7736447307554661,
      "compression_loss": 0.0,
      "distillation_loss": 0.073527991771698,
      "epoch": 2.87,
      "learning_rate": 3.198613457339493e-05,
      "loss": 0.0798,
      "step": 3024,
      "task_loss": 0.13668204843997955
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7736593671504914,
      "compression_loss": 0.0,
      "distillation_loss": 0.017919428646564484,
      "epoch": 2.87,
      "learning_rate": 3.19759031814872e-05,
      "loss": 0.0165,
      "step": 3025,
      "task_loss": 0.0033756643533706665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7736739981256334,
      "compression_loss": 0.0,
      "distillation_loss": 0.03155189007520676,
      "epoch": 2.87,
      "learning_rate": 3.196567052237306e-05,
      "loss": 0.0288,
      "step": 3026,
      "task_loss": 0.003785140812397003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7736886236818958,
      "compression_loss": 0.0,
      "distillation_loss": 0.10638611018657684,
      "epoch": 2.87,
      "learning_rate": 3.195543659791132e-05,
      "loss": 0.1069,
      "step": 3027,
      "task_loss": 0.11161436140537262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7737032438202823,
      "compression_loss": 0.0,
      "distillation_loss": 0.19940927624702454,
      "epoch": 2.88,
      "learning_rate": 3.194520140996102e-05,
      "loss": 0.1983,
      "step": 3028,
      "task_loss": 0.1886705756187439
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7737178585417965,
      "compression_loss": 0.0,
      "distillation_loss": 0.0578513965010643,
      "epoch": 2.88,
      "learning_rate": 3.193496496038144e-05,
      "loss": 0.0666,
      "step": 3029,
      "task_loss": 0.1448466032743454
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7737324678474422,
      "compression_loss": 0.0,
      "distillation_loss": 0.18462207913398743,
      "epoch": 2.88,
      "learning_rate": 3.1924727251032075e-05,
      "loss": 0.1751,
      "step": 3030,
      "task_loss": 0.08917995542287827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7737470717382229,
      "compression_loss": 0.0,
      "distillation_loss": 0.0979708880186081,
      "epoch": 2.88,
      "learning_rate": 3.191448828377267e-05,
      "loss": 0.1038,
      "step": 3031,
      "task_loss": 0.15621528029441833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7737616702151424,
      "compression_loss": 0.0,
      "distillation_loss": 0.06346909701824188,
      "epoch": 2.88,
      "learning_rate": 3.1904248060463146e-05,
      "loss": 0.0632,
      "step": 3032,
      "task_loss": 0.06106296554207802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7737762632792045,
      "compression_loss": 0.0,
      "distillation_loss": 0.08692383021116257,
      "epoch": 2.88,
      "learning_rate": 3.189400658296372e-05,
      "loss": 0.0864,
      "step": 3033,
      "task_loss": 0.08201731741428375
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7737908509314126,
      "compression_loss": 0.0,
      "distillation_loss": 0.06603842228651047,
      "epoch": 2.88,
      "learning_rate": 3.188376385313479e-05,
      "loss": 0.0611,
      "step": 3034,
      "task_loss": 0.016200372949242592
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7738054331727706,
      "compression_loss": 0.0,
      "distillation_loss": 0.1842041015625,
      "epoch": 2.88,
      "learning_rate": 3.187351987283701e-05,
      "loss": 0.1758,
      "step": 3035,
      "task_loss": 0.10015082359313965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7738200100042821,
      "compression_loss": 0.0,
      "distillation_loss": 0.1197400689125061,
      "epoch": 2.88,
      "learning_rate": 3.1863274643931244e-05,
      "loss": 0.1273,
      "step": 3036,
      "task_loss": 0.19555895030498505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7738345814269508,
      "compression_loss": 0.0,
      "distillation_loss": 0.24975571036338806,
      "epoch": 2.88,
      "learning_rate": 3.185302816827858e-05,
      "loss": 0.242,
      "step": 3037,
      "task_loss": 0.17197510600090027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7738491474417805,
      "compression_loss": 0.0,
      "distillation_loss": 0.0718345120549202,
      "epoch": 2.89,
      "learning_rate": 3.184278044774035e-05,
      "loss": 0.0706,
      "step": 3038,
      "task_loss": 0.0589919239282608
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7738637080497746,
      "compression_loss": 0.0,
      "distillation_loss": 0.17170128226280212,
      "epoch": 2.89,
      "learning_rate": 3.183253148417808e-05,
      "loss": 0.1797,
      "step": 3039,
      "task_loss": 0.25216758251190186
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.773878263251937,
      "compression_loss": 0.0,
      "distillation_loss": 0.17050856351852417,
      "epoch": 2.89,
      "learning_rate": 3.182228127945358e-05,
      "loss": 0.1782,
      "step": 3040,
      "task_loss": 0.2472606599330902
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7738928130492714,
      "compression_loss": 0.0,
      "distillation_loss": 0.09137849509716034,
      "epoch": 2.89,
      "learning_rate": 3.1812029835428825e-05,
      "loss": 0.0876,
      "step": 3041,
      "task_loss": 0.053437259048223495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7739073574427813,
      "compression_loss": 0.0,
      "distillation_loss": 0.08643493056297302,
      "epoch": 2.89,
      "learning_rate": 3.1801777153966034e-05,
      "loss": 0.1166,
      "step": 3042,
      "task_loss": 0.3879551291465759
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7739218964334705,
      "compression_loss": 0.0,
      "distillation_loss": 0.06834743916988373,
      "epoch": 2.89,
      "learning_rate": 3.179152323692767e-05,
      "loss": 0.0775,
      "step": 3043,
      "task_loss": 0.16033974289894104
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7739364300223428,
      "compression_loss": 0.0,
      "distillation_loss": 0.07843412458896637,
      "epoch": 2.89,
      "learning_rate": 3.1781268086176406e-05,
      "loss": 0.0788,
      "step": 3044,
      "task_loss": 0.08182865381240845
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7739509582104016,
      "compression_loss": 0.0,
      "distillation_loss": 0.09358078241348267,
      "epoch": 2.89,
      "learning_rate": 3.177101170357513e-05,
      "loss": 0.0883,
      "step": 3045,
      "task_loss": 0.04088345542550087
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7739654809986508,
      "compression_loss": 0.0,
      "distillation_loss": 0.030289195477962494,
      "epoch": 2.89,
      "learning_rate": 3.1760754090986975e-05,
      "loss": 0.028,
      "step": 3046,
      "task_loss": 0.007276715710759163
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7739799983880941,
      "compression_loss": 0.0,
      "distillation_loss": 0.06697987020015717,
      "epoch": 2.89,
      "learning_rate": 3.175049525027527e-05,
      "loss": 0.0772,
      "step": 3047,
      "task_loss": 0.16944551467895508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.773994510379735,
      "compression_loss": 0.0,
      "distillation_loss": 0.09630005806684494,
      "epoch": 2.89,
      "learning_rate": 3.17402351833036e-05,
      "loss": 0.0926,
      "step": 3048,
      "task_loss": 0.059294380247592926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7740090169745774,
      "compression_loss": 0.0,
      "distillation_loss": 0.04504215717315674,
      "epoch": 2.9,
      "learning_rate": 3.1729973891935745e-05,
      "loss": 0.0579,
      "step": 3049,
      "task_loss": 0.17366653680801392
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7740235181736247,
      "compression_loss": 0.0,
      "distillation_loss": 0.024151597172021866,
      "epoch": 2.9,
      "learning_rate": 3.1719711378035714e-05,
      "loss": 0.0223,
      "step": 3050,
      "task_loss": 0.005144516006112099
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7740380139778809,
      "compression_loss": 0.0,
      "distillation_loss": 0.07286585122346878,
      "epoch": 2.9,
      "learning_rate": 3.1709447643467755e-05,
      "loss": 0.0733,
      "step": 3051,
      "task_loss": 0.07735887914896011
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7740525043883495,
      "compression_loss": 0.0,
      "distillation_loss": 0.12737995386123657,
      "epoch": 2.9,
      "learning_rate": 3.1699182690096316e-05,
      "loss": 0.1227,
      "step": 3052,
      "task_loss": 0.08107961714267731
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7740669894060341,
      "compression_loss": 0.0,
      "distillation_loss": 0.13144919276237488,
      "epoch": 2.9,
      "learning_rate": 3.168891651978609e-05,
      "loss": 0.1325,
      "step": 3053,
      "task_loss": 0.14158479869365692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7740814690319386,
      "compression_loss": 0.0,
      "distillation_loss": 0.04931913688778877,
      "epoch": 2.9,
      "learning_rate": 3.167864913440195e-05,
      "loss": 0.0455,
      "step": 3054,
      "task_loss": 0.011026356369256973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7740959432670665,
      "compression_loss": 0.0,
      "distillation_loss": 0.032145917415618896,
      "epoch": 2.9,
      "learning_rate": 3.1668380535809036e-05,
      "loss": 0.0341,
      "step": 3055,
      "task_loss": 0.05174173414707184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7741104121124217,
      "compression_loss": 0.0,
      "distillation_loss": 0.04694236442446709,
      "epoch": 2.9,
      "learning_rate": 3.165811072587267e-05,
      "loss": 0.0486,
      "step": 3056,
      "task_loss": 0.06309117376804352
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7741248755690077,
      "compression_loss": 0.0,
      "distillation_loss": 0.13567635416984558,
      "epoch": 2.9,
      "learning_rate": 3.164783970645845e-05,
      "loss": 0.1325,
      "step": 3057,
      "task_loss": 0.10343907028436661
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7741393336378282,
      "compression_loss": 0.0,
      "distillation_loss": 0.11043036729097366,
      "epoch": 2.9,
      "learning_rate": 3.1637567479432113e-05,
      "loss": 0.1153,
      "step": 3058,
      "task_loss": 0.15953701734542847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7741537863198868,
      "compression_loss": 0.0,
      "distillation_loss": 0.06733449548482895,
      "epoch": 2.91,
      "learning_rate": 3.1627294046659675e-05,
      "loss": 0.0654,
      "step": 3059,
      "task_loss": 0.04814068600535393
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7741682336161875,
      "compression_loss": 0.0,
      "distillation_loss": 0.030272886157035828,
      "epoch": 2.91,
      "learning_rate": 3.1617019410007366e-05,
      "loss": 0.0279,
      "step": 3060,
      "task_loss": 0.006075674667954445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7741826755277337,
      "compression_loss": 0.0,
      "distillation_loss": 0.16811983287334442,
      "epoch": 2.91,
      "learning_rate": 3.16067435713416e-05,
      "loss": 0.1535,
      "step": 3061,
      "task_loss": 0.021725336089730263
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.774197112055529,
      "compression_loss": 0.0,
      "distillation_loss": 0.024632154032588005,
      "epoch": 2.91,
      "learning_rate": 3.159646653252906e-05,
      "loss": 0.0229,
      "step": 3062,
      "task_loss": 0.007006222382187843
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7742115432005774,
      "compression_loss": 0.0,
      "distillation_loss": 0.19798026978969574,
      "epoch": 2.91,
      "learning_rate": 3.1586188295436594e-05,
      "loss": 0.1975,
      "step": 3063,
      "task_loss": 0.19281329214572906
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7742259689638823,
      "compression_loss": 0.0,
      "distillation_loss": 0.08180099725723267,
      "epoch": 2.91,
      "learning_rate": 3.1575908861931314e-05,
      "loss": 0.0904,
      "step": 3064,
      "task_loss": 0.16796118021011353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7742403893464476,
      "compression_loss": 0.0,
      "distillation_loss": 0.0645010769367218,
      "epoch": 2.91,
      "learning_rate": 3.156562823388051e-05,
      "loss": 0.0614,
      "step": 3065,
      "task_loss": 0.03324957937002182
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7742548043492768,
      "compression_loss": 0.0,
      "distillation_loss": 0.10893145948648453,
      "epoch": 2.91,
      "learning_rate": 3.155534641315172e-05,
      "loss": 0.1009,
      "step": 3066,
      "task_loss": 0.028400206938385963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7742692139733738,
      "compression_loss": 0.0,
      "distillation_loss": 0.032349854707717896,
      "epoch": 2.91,
      "learning_rate": 3.154506340161269e-05,
      "loss": 0.0305,
      "step": 3067,
      "task_loss": 0.013784002512693405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.774283618219742,
      "compression_loss": 0.0,
      "distillation_loss": 0.19412454962730408,
      "epoch": 2.91,
      "learning_rate": 3.1534779201131366e-05,
      "loss": 0.1937,
      "step": 3068,
      "task_loss": 0.1897927224636078
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7742980170893853,
      "compression_loss": 0.0,
      "distillation_loss": 0.02269315905869007,
      "epoch": 2.91,
      "learning_rate": 3.152449381357593e-05,
      "loss": 0.021,
      "step": 3069,
      "task_loss": 0.005848288536071777
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7743124105833072,
      "compression_loss": 0.0,
      "distillation_loss": 0.17387601733207703,
      "epoch": 2.92,
      "learning_rate": 3.151420724081478e-05,
      "loss": 0.1707,
      "step": 3070,
      "task_loss": 0.14171260595321655
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7743267987025116,
      "compression_loss": 0.0,
      "distillation_loss": 0.0965137854218483,
      "epoch": 2.92,
      "learning_rate": 3.1503919484716495e-05,
      "loss": 0.0902,
      "step": 3071,
      "task_loss": 0.03351738303899765
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7743411814480021,
      "compression_loss": 0.0,
      "distillation_loss": 0.05635223537683487,
      "epoch": 2.92,
      "learning_rate": 3.149363054714992e-05,
      "loss": 0.056,
      "step": 3072,
      "task_loss": 0.05248015746474266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7743555588207823,
      "compression_loss": 0.0,
      "distillation_loss": 0.12340125441551208,
      "epoch": 2.92,
      "learning_rate": 3.148334042998408e-05,
      "loss": 0.1179,
      "step": 3073,
      "task_loss": 0.06886960566043854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.774369930821856,
      "compression_loss": 0.0,
      "distillation_loss": 0.06106632202863693,
      "epoch": 2.92,
      "learning_rate": 3.1473049135088225e-05,
      "loss": 0.0575,
      "step": 3074,
      "task_loss": 0.025458548218011856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7743842974522268,
      "compression_loss": 0.0,
      "distillation_loss": 0.04669322818517685,
      "epoch": 2.92,
      "learning_rate": 3.146275666433183e-05,
      "loss": 0.0486,
      "step": 3075,
      "task_loss": 0.06542003899812698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7743986587128984,
      "compression_loss": 0.0,
      "distillation_loss": 0.15752683579921722,
      "epoch": 2.92,
      "learning_rate": 3.145246301958455e-05,
      "loss": 0.1552,
      "step": 3076,
      "task_loss": 0.13474304974079132
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7744130146048744,
      "compression_loss": 0.0,
      "distillation_loss": 0.040889352560043335,
      "epoch": 2.92,
      "learning_rate": 3.14421682027163e-05,
      "loss": 0.0456,
      "step": 3077,
      "task_loss": 0.08812226355075836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7744273651291587,
      "compression_loss": 0.0,
      "distillation_loss": 0.07757270336151123,
      "epoch": 2.92,
      "learning_rate": 3.143187221559715e-05,
      "loss": 0.0816,
      "step": 3078,
      "task_loss": 0.11832978576421738
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7744417102867547,
      "compression_loss": 0.0,
      "distillation_loss": 0.04706088453531265,
      "epoch": 2.92,
      "learning_rate": 3.142157506009746e-05,
      "loss": 0.0517,
      "step": 3079,
      "task_loss": 0.0932609811425209
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7744560500786662,
      "compression_loss": 0.0,
      "distillation_loss": 0.17205679416656494,
      "epoch": 2.92,
      "learning_rate": 3.141127673808772e-05,
      "loss": 0.1766,
      "step": 3080,
      "task_loss": 0.21797949075698853
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7744703845058971,
      "compression_loss": 0.0,
      "distillation_loss": 0.11091163754463196,
      "epoch": 2.93,
      "learning_rate": 3.140097725143868e-05,
      "loss": 0.1031,
      "step": 3081,
      "task_loss": 0.03264191001653671
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7744847135694508,
      "compression_loss": 0.0,
      "distillation_loss": 0.17644111812114716,
      "epoch": 2.93,
      "learning_rate": 3.13906766020213e-05,
      "loss": 0.1904,
      "step": 3082,
      "task_loss": 0.3157804012298584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7744990372703311,
      "compression_loss": 0.0,
      "distillation_loss": 0.037624701857566833,
      "epoch": 2.93,
      "learning_rate": 3.138037479170674e-05,
      "loss": 0.0394,
      "step": 3083,
      "task_loss": 0.05507116764783859
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7745133556095416,
      "compression_loss": 0.0,
      "distillation_loss": 0.14321953058242798,
      "epoch": 2.93,
      "learning_rate": 3.137007182236637e-05,
      "loss": 0.1347,
      "step": 3084,
      "task_loss": 0.05840989202260971
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7745276685880862,
      "compression_loss": 0.0,
      "distillation_loss": 0.03971727564930916,
      "epoch": 2.93,
      "learning_rate": 3.1359767695871767e-05,
      "loss": 0.0431,
      "step": 3085,
      "task_loss": 0.07382907718420029
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7745419762069683,
      "compression_loss": 0.0,
      "distillation_loss": 0.3982169032096863,
      "epoch": 2.93,
      "learning_rate": 3.134946241409474e-05,
      "loss": 0.3973,
      "step": 3086,
      "task_loss": 0.3887830376625061
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7745562784671917,
      "compression_loss": 0.0,
      "distillation_loss": 0.03360104188323021,
      "epoch": 2.93,
      "learning_rate": 3.133915597890729e-05,
      "loss": 0.0468,
      "step": 3087,
      "task_loss": 0.1660730242729187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.77457057536976,
      "compression_loss": 0.0,
      "distillation_loss": 0.030526025220751762,
      "epoch": 2.93,
      "learning_rate": 3.132884839218162e-05,
      "loss": 0.0331,
      "step": 3088,
      "task_loss": 0.056509580463171005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7745848669156772,
      "compression_loss": 0.0,
      "distillation_loss": 0.02792483940720558,
      "epoch": 2.93,
      "learning_rate": 3.131853965579016e-05,
      "loss": 0.0285,
      "step": 3089,
      "task_loss": 0.03412621468305588
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7745991531059465,
      "compression_loss": 0.0,
      "distillation_loss": 0.10251462459564209,
      "epoch": 2.93,
      "learning_rate": 3.130822977160554e-05,
      "loss": 0.0957,
      "step": 3090,
      "task_loss": 0.03391581028699875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.774613433941572,
      "compression_loss": 0.0,
      "distillation_loss": 0.04014907404780388,
      "epoch": 2.94,
      "learning_rate": 3.129791874150062e-05,
      "loss": 0.0426,
      "step": 3091,
      "task_loss": 0.0648120641708374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7746277094235572,
      "compression_loss": 0.0,
      "distillation_loss": 0.0542302206158638,
      "epoch": 2.94,
      "learning_rate": 3.1287606567348406e-05,
      "loss": 0.0595,
      "step": 3092,
      "task_loss": 0.10736650973558426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7746419795529058,
      "compression_loss": 0.0,
      "distillation_loss": 0.02264413796365261,
      "epoch": 2.94,
      "learning_rate": 3.1277293251022185e-05,
      "loss": 0.0265,
      "step": 3093,
      "task_loss": 0.0615130253136158
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7746562443306214,
      "compression_loss": 0.0,
      "distillation_loss": 0.16324403882026672,
      "epoch": 2.94,
      "learning_rate": 3.126697879439541e-05,
      "loss": 0.1576,
      "step": 3094,
      "task_loss": 0.10708633810281754
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7746705037577079,
      "compression_loss": 0.0,
      "distillation_loss": 0.025752229616045952,
      "epoch": 2.94,
      "learning_rate": 3.1256663199341764e-05,
      "loss": 0.0244,
      "step": 3095,
      "task_loss": 0.012307420372962952
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7746847578351687,
      "compression_loss": 0.0,
      "distillation_loss": 0.02341257967054844,
      "epoch": 2.94,
      "learning_rate": 3.124634646773511e-05,
      "loss": 0.0219,
      "step": 3096,
      "task_loss": 0.008523575961589813
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7746990065640077,
      "compression_loss": 0.0,
      "distillation_loss": 0.11951486021280289,
      "epoch": 2.94,
      "learning_rate": 3.1236028601449534e-05,
      "loss": 0.1255,
      "step": 3097,
      "task_loss": 0.17937280237674713
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7747132499452285,
      "compression_loss": 0.0,
      "distillation_loss": 0.1342199444770813,
      "epoch": 2.94,
      "learning_rate": 3.1225709602359335e-05,
      "loss": 0.1285,
      "step": 3098,
      "task_loss": 0.07676542550325394
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7747274879798347,
      "compression_loss": 0.0,
      "distillation_loss": 0.04866989701986313,
      "epoch": 2.94,
      "learning_rate": 3.1215389472338995e-05,
      "loss": 0.0541,
      "step": 3099,
      "task_loss": 0.10313687473535538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7747417206688302,
      "compression_loss": 0.0,
      "distillation_loss": 0.09249399602413177,
      "epoch": 2.94,
      "learning_rate": 3.1205068213263234e-05,
      "loss": 0.1012,
      "step": 3100,
      "task_loss": 0.17944881319999695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7747559480132186,
      "compression_loss": 0.0,
      "distillation_loss": 0.05482405051589012,
      "epoch": 2.94,
      "learning_rate": 3.119474582700695e-05,
      "loss": 0.0604,
      "step": 3101,
      "task_loss": 0.11096543818712234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7747701700140034,
      "compression_loss": 0.0,
      "distillation_loss": 0.02817433886229992,
      "epoch": 2.95,
      "learning_rate": 3.118442231544524e-05,
      "loss": 0.0331,
      "step": 3102,
      "task_loss": 0.07782773673534393
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7747843866721884,
      "compression_loss": 0.0,
      "distillation_loss": 0.03213953226804733,
      "epoch": 2.95,
      "learning_rate": 3.117409768045344e-05,
      "loss": 0.0392,
      "step": 3103,
      "task_loss": 0.10243944823741913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7747985979887774,
      "compression_loss": 0.0,
      "distillation_loss": 0.2645414471626282,
      "epoch": 2.95,
      "learning_rate": 3.116377192390706e-05,
      "loss": 0.2525,
      "step": 3104,
      "task_loss": 0.14380747079849243
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7748128039647739,
      "compression_loss": 0.0,
      "distillation_loss": 0.03608560562133789,
      "epoch": 2.95,
      "learning_rate": 3.115344504768183e-05,
      "loss": 0.044,
      "step": 3105,
      "task_loss": 0.115634024143219
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7748270046011818,
      "compression_loss": 0.0,
      "distillation_loss": 0.0855831503868103,
      "epoch": 2.95,
      "learning_rate": 3.1143117053653665e-05,
      "loss": 0.0873,
      "step": 3106,
      "task_loss": 0.10302843898534775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7748411998990046,
      "compression_loss": 0.0,
      "distillation_loss": 0.1709257960319519,
      "epoch": 2.95,
      "learning_rate": 3.113278794369869e-05,
      "loss": 0.1625,
      "step": 3107,
      "task_loss": 0.08654191344976425
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.774855389859246,
      "compression_loss": 0.0,
      "distillation_loss": 0.012697904370725155,
      "epoch": 2.95,
      "learning_rate": 3.112245771969327e-05,
      "loss": 0.0119,
      "step": 3108,
      "task_loss": 0.004779975861310959
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7748695744829096,
      "compression_loss": 0.0,
      "distillation_loss": 0.12192719429731369,
      "epoch": 2.95,
      "learning_rate": 3.1112126383513914e-05,
      "loss": 0.1268,
      "step": 3109,
      "task_loss": 0.1710333675146103
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7748837537709993,
      "compression_loss": 0.0,
      "distillation_loss": 0.09610234200954437,
      "epoch": 2.95,
      "learning_rate": 3.110179393703737e-05,
      "loss": 0.1014,
      "step": 3110,
      "task_loss": 0.1487446278333664
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7748979277245187,
      "compression_loss": 0.0,
      "distillation_loss": 0.12404096126556396,
      "epoch": 2.95,
      "learning_rate": 3.109146038214055e-05,
      "loss": 0.1215,
      "step": 3111,
      "task_loss": 0.09883137792348862
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7749120963444714,
      "compression_loss": 0.0,
      "distillation_loss": 0.07528949528932571,
      "epoch": 2.96,
      "learning_rate": 3.108112572070063e-05,
      "loss": 0.0717,
      "step": 3112,
      "task_loss": 0.03988656401634216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7749262596318612,
      "compression_loss": 0.0,
      "distillation_loss": 0.06444555521011353,
      "epoch": 2.96,
      "learning_rate": 3.1070789954594934e-05,
      "loss": 0.0607,
      "step": 3113,
      "task_loss": 0.027471771463751793
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7749404175876917,
      "compression_loss": 0.0,
      "distillation_loss": 0.06135058403015137,
      "epoch": 2.96,
      "learning_rate": 3.1060453085701e-05,
      "loss": 0.0668,
      "step": 3114,
      "task_loss": 0.11571139842271805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7749545702129665,
      "compression_loss": 0.0,
      "distillation_loss": 0.15376602113246918,
      "epoch": 2.96,
      "learning_rate": 3.105011511589658e-05,
      "loss": 0.1476,
      "step": 3115,
      "task_loss": 0.09177757799625397
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7749687175086896,
      "compression_loss": 0.0,
      "distillation_loss": 0.04827346280217171,
      "epoch": 2.96,
      "learning_rate": 3.103977604705961e-05,
      "loss": 0.059,
      "step": 3116,
      "task_loss": 0.15540531277656555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7749828594758643,
      "compression_loss": 0.0,
      "distillation_loss": 0.08922263234853745,
      "epoch": 2.96,
      "learning_rate": 3.102943588106824e-05,
      "loss": 0.0912,
      "step": 3117,
      "task_loss": 0.10851528495550156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7749969961154946,
      "compression_loss": 0.0,
      "distillation_loss": 0.03960554301738739,
      "epoch": 2.96,
      "learning_rate": 3.10190946198008e-05,
      "loss": 0.0461,
      "step": 3118,
      "task_loss": 0.10426989942789078
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7750111274285838,
      "compression_loss": 0.0,
      "distillation_loss": 0.05592392012476921,
      "epoch": 2.96,
      "learning_rate": 3.100875226513583e-05,
      "loss": 0.0597,
      "step": 3119,
      "task_loss": 0.09348025172948837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7750252534161359,
      "compression_loss": 0.0,
      "distillation_loss": 0.05171472951769829,
      "epoch": 2.96,
      "learning_rate": 3.099840881895208e-05,
      "loss": 0.0476,
      "step": 3120,
      "task_loss": 0.010983137413859367
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7750393740791546,
      "compression_loss": 0.0,
      "distillation_loss": 0.09998922049999237,
      "epoch": 2.96,
      "learning_rate": 3.098806428312847e-05,
      "loss": 0.0984,
      "step": 3121,
      "task_loss": 0.08365054428577423
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7750534894186435,
      "compression_loss": 0.0,
      "distillation_loss": 0.020870694890618324,
      "epoch": 2.96,
      "learning_rate": 3.097771865954415e-05,
      "loss": 0.0196,
      "step": 3122,
      "task_loss": 0.008448204025626183
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7750675994356061,
      "compression_loss": 0.0,
      "distillation_loss": 0.1635720431804657,
      "epoch": 2.97,
      "learning_rate": 3.096737195007845e-05,
      "loss": 0.1693,
      "step": 3123,
      "task_loss": 0.22086140513420105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7750817041310464,
      "compression_loss": 0.0,
      "distillation_loss": 0.23271842300891876,
      "epoch": 2.97,
      "learning_rate": 3.0957024156610884e-05,
      "loss": 0.2299,
      "step": 3124,
      "task_loss": 0.20500387251377106
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7750958035059679,
      "compression_loss": 0.0,
      "distillation_loss": 0.0583171546459198,
      "epoch": 2.97,
      "learning_rate": 3.09466752810212e-05,
      "loss": 0.0556,
      "step": 3125,
      "task_loss": 0.03121706284582615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7751098975613742,
      "compression_loss": 0.0,
      "distillation_loss": 0.06426501274108887,
      "epoch": 2.97,
      "learning_rate": 3.093632532518931e-05,
      "loss": 0.0598,
      "step": 3126,
      "task_loss": 0.019734475761651993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7751239862982691,
      "compression_loss": 0.0,
      "distillation_loss": 0.15301446616649628,
      "epoch": 2.97,
      "learning_rate": 3.092597429099534e-05,
      "loss": 0.1451,
      "step": 3127,
      "task_loss": 0.07339885830879211
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7751380697176563,
      "compression_loss": 0.0,
      "distillation_loss": 0.10157543420791626,
      "epoch": 2.97,
      "learning_rate": 3.0915622180319585e-05,
      "loss": 0.1026,
      "step": 3128,
      "task_loss": 0.11221370846033096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7751521478205394,
      "compression_loss": 0.0,
      "distillation_loss": 0.10004980117082596,
      "epoch": 2.97,
      "learning_rate": 3.090526899504259e-05,
      "loss": 0.096,
      "step": 3129,
      "task_loss": 0.05929484963417053
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7751662206079223,
      "compression_loss": 0.0,
      "distillation_loss": 0.035803135484457016,
      "epoch": 2.97,
      "learning_rate": 3.0894914737045034e-05,
      "loss": 0.0331,
      "step": 3130,
      "task_loss": 0.009108470752835274
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7751802880808084,
      "compression_loss": 0.0,
      "distillation_loss": 0.12775352597236633,
      "epoch": 2.97,
      "learning_rate": 3.088455940820782e-05,
      "loss": 0.1286,
      "step": 3131,
      "task_loss": 0.1366499364376068
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7751943502402016,
      "compression_loss": 0.0,
      "distillation_loss": 0.19691312313079834,
      "epoch": 2.97,
      "learning_rate": 3.087420301041206e-05,
      "loss": 0.1927,
      "step": 3132,
      "task_loss": 0.15475985407829285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7752084070871054,
      "compression_loss": 0.0,
      "distillation_loss": 0.19966569542884827,
      "epoch": 2.98,
      "learning_rate": 3.086384554553902e-05,
      "loss": 0.1948,
      "step": 3133,
      "task_loss": 0.1510712206363678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7752224586225235,
      "compression_loss": 0.0,
      "distillation_loss": 0.13692443072795868,
      "epoch": 2.98,
      "learning_rate": 3.0853487015470206e-05,
      "loss": 0.1358,
      "step": 3134,
      "task_loss": 0.12599924206733704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7752365048474598,
      "compression_loss": 0.0,
      "distillation_loss": 0.23906008899211884,
      "epoch": 2.98,
      "learning_rate": 3.084312742208728e-05,
      "loss": 0.2498,
      "step": 3135,
      "task_loss": 0.34651899337768555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7752505457629176,
      "compression_loss": 0.0,
      "distillation_loss": 0.03585006296634674,
      "epoch": 2.98,
      "learning_rate": 3.083276676727212e-05,
      "loss": 0.034,
      "step": 3136,
      "task_loss": 0.016858315095305443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7752645813699011,
      "compression_loss": 0.0,
      "distillation_loss": 0.10707780718803406,
      "epoch": 2.98,
      "learning_rate": 3.082240505290678e-05,
      "loss": 0.1068,
      "step": 3137,
      "task_loss": 0.10397090762853622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7752786116694135,
      "compression_loss": 0.0,
      "distillation_loss": 0.09790733456611633,
      "epoch": 2.98,
      "learning_rate": 3.081204228087353e-05,
      "loss": 0.091,
      "step": 3138,
      "task_loss": 0.029314137995243073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7752926366624586,
      "compression_loss": 0.0,
      "distillation_loss": 0.298001766204834,
      "epoch": 2.98,
      "learning_rate": 3.08016784530548e-05,
      "loss": 0.2974,
      "step": 3139,
      "task_loss": 0.2921411693096161
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7753066563500404,
      "compression_loss": 0.0,
      "distillation_loss": 0.13080675899982452,
      "epoch": 2.98,
      "learning_rate": 3.0791313571333244e-05,
      "loss": 0.1232,
      "step": 3140,
      "task_loss": 0.05510927364230156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7753206707331621,
      "compression_loss": 0.0,
      "distillation_loss": 0.10794495791196823,
      "epoch": 2.98,
      "learning_rate": 3.078094763759168e-05,
      "loss": 0.1087,
      "step": 3141,
      "task_loss": 0.1155327633023262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7753346798128278,
      "compression_loss": 0.0,
      "distillation_loss": 0.0587034747004509,
      "epoch": 2.98,
      "learning_rate": 3.0770580653713146e-05,
      "loss": 0.0565,
      "step": 3142,
      "task_loss": 0.03710935637354851
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7753486835900409,
      "compression_loss": 0.0,
      "distillation_loss": 0.09703940898180008,
      "epoch": 2.98,
      "learning_rate": 3.076021262158084e-05,
      "loss": 0.1103,
      "step": 3143,
      "task_loss": 0.22997155785560608
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7753626820658053,
      "compression_loss": 0.0,
      "distillation_loss": 0.0734797865152359,
      "epoch": 2.99,
      "learning_rate": 3.074984354307817e-05,
      "loss": 0.0786,
      "step": 3144,
      "task_loss": 0.1250114142894745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7753766752411244,
      "compression_loss": 0.0,
      "distillation_loss": 0.10003338009119034,
      "epoch": 2.99,
      "learning_rate": 3.073947342008873e-05,
      "loss": 0.1057,
      "step": 3145,
      "task_loss": 0.15664803981781006
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7753906631170021,
      "compression_loss": 0.0,
      "distillation_loss": 0.11784384399652481,
      "epoch": 2.99,
      "learning_rate": 3.07291022544963e-05,
      "loss": 0.124,
      "step": 3146,
      "task_loss": 0.17987778782844543
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7754046456944421,
      "compression_loss": 0.0,
      "distillation_loss": 0.14313513040542603,
      "epoch": 2.99,
      "learning_rate": 3.0718730048184855e-05,
      "loss": 0.1502,
      "step": 3147,
      "task_loss": 0.2141174077987671
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.775418622974448,
      "compression_loss": 0.0,
      "distillation_loss": 0.027550848200917244,
      "epoch": 2.99,
      "learning_rate": 3.0708356803038556e-05,
      "loss": 0.0306,
      "step": 3148,
      "task_loss": 0.05766863748431206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7754325949580235,
      "compression_loss": 0.0,
      "distillation_loss": 0.11986173689365387,
      "epoch": 2.99,
      "learning_rate": 3.069798252094175e-05,
      "loss": 0.1198,
      "step": 3149,
      "task_loss": 0.11939980834722519
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7754465616461722,
      "compression_loss": 0.0,
      "distillation_loss": 0.05384514853358269,
      "epoch": 2.99,
      "learning_rate": 3.068760720377897e-05,
      "loss": 0.0502,
      "step": 3150,
      "task_loss": 0.017584798857569695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7754605230398979,
      "compression_loss": 0.0,
      "distillation_loss": 0.12095589935779572,
      "epoch": 2.99,
      "learning_rate": 3.067723085343496e-05,
      "loss": 0.1254,
      "step": 3151,
      "task_loss": 0.16499733924865723
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7754744791402043,
      "compression_loss": 0.0,
      "distillation_loss": 0.23978760838508606,
      "epoch": 2.99,
      "learning_rate": 3.066685347179462e-05,
      "loss": 0.2298,
      "step": 3152,
      "task_loss": 0.1398588865995407
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.775488429948095,
      "compression_loss": 0.0,
      "distillation_loss": 0.016089707612991333,
      "epoch": 2.99,
      "learning_rate": 3.065647506074306e-05,
      "loss": 0.0148,
      "step": 3153,
      "task_loss": 0.002878313884139061
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7755023754645736,
      "compression_loss": 0.0,
      "distillation_loss": 0.13207751512527466,
      "epoch": 3.0,
      "learning_rate": 3.064609562216555e-05,
      "loss": 0.1289,
      "step": 3154,
      "task_loss": 0.10021229833364487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7755163156906439,
      "compression_loss": 0.0,
      "distillation_loss": 0.022115526720881462,
      "epoch": 3.0,
      "learning_rate": 3.063571515794759e-05,
      "loss": 0.0208,
      "step": 3155,
      "task_loss": 0.008827542886137962
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7755302506273097,
      "compression_loss": 0.0,
      "distillation_loss": 0.07137614488601685,
      "epoch": 3.0,
      "learning_rate": 3.062533366997483e-05,
      "loss": 0.0964,
      "step": 3156,
      "task_loss": 0.3213070034980774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7755441802755745,
      "compression_loss": 0.0,
      "distillation_loss": 0.01968713104724884,
      "epoch": 3.0,
      "learning_rate": 3.061495116013311e-05,
      "loss": 0.0209,
      "step": 3157,
      "task_loss": 0.03145528584718704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.775558104636442,
      "compression_loss": 0.0,
      "distillation_loss": 0.02357085794210434,
      "epoch": 3.0,
      "learning_rate": 3.060456763030847e-05,
      "loss": 0.0219,
      "step": 3158,
      "task_loss": 0.006720980629324913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.5919843851205062,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.757812535077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7755720237109159,
      "compression_loss": 0.0,
      "distillation_loss": 0.11487483233213425,
      "epoch": 3.0,
      "learning_rate": 3.059418308238713e-05,
      "loss": 0.1098,
      "step": 3159,
      "task_loss": 0.06400777399539948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7755859375,
      "compression_loss": 0.0,
      "distillation_loss": 0.3807898461818695,
      "epoch": 3.0,
      "learning_rate": 3.0583797518255505e-05,
      "loss": 0.3608,
      "step": 3160,
      "task_loss": 0.18057574331760406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7755998460046979,
      "compression_loss": 0.0,
      "distillation_loss": 0.08554988354444504,
      "epoch": 3.0,
      "learning_rate": 3.057341093980015e-05,
      "loss": 0.081,
      "step": 3161,
      "task_loss": 0.04051545634865761
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7756137492260131,
      "compression_loss": 0.0,
      "distillation_loss": 0.5044205784797668,
      "epoch": 3.0,
      "learning_rate": 3.056302334890786e-05,
      "loss": 0.4752,
      "step": 3162,
      "task_loss": 0.2119530588388443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7756276471649496,
      "compression_loss": 0.0,
      "distillation_loss": 0.7463757395744324,
      "epoch": 3.0,
      "learning_rate": 3.055263474746559e-05,
      "loss": 0.7219,
      "step": 3163,
      "task_loss": 0.5015002489089966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7756415398225108,
      "compression_loss": 0.0,
      "distillation_loss": 0.15632954239845276,
      "epoch": 3.0,
      "learning_rate": 3.054224513736048e-05,
      "loss": 0.1498,
      "step": 3164,
      "task_loss": 0.09135162830352783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7756554271997006,
      "compression_loss": 0.0,
      "distillation_loss": 0.4045957326889038,
      "epoch": 3.01,
      "learning_rate": 3.0531854520479844e-05,
      "loss": 0.3806,
      "step": 3165,
      "task_loss": 0.16499730944633484
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7756693092975225,
      "compression_loss": 0.0,
      "distillation_loss": 0.43263140320777893,
      "epoch": 3.01,
      "learning_rate": 3.05214628987112e-05,
      "loss": 0.4179,
      "step": 3166,
      "task_loss": 0.2850673496723175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7756831861169803,
      "compression_loss": 0.0,
      "distillation_loss": 0.23246167600154877,
      "epoch": 3.01,
      "learning_rate": 3.0511070273942217e-05,
      "loss": 0.2326,
      "step": 3167,
      "task_loss": 0.23360560834407806
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7756970576590777,
      "compression_loss": 0.0,
      "distillation_loss": 0.1095157265663147,
      "epoch": 3.01,
      "learning_rate": 3.0500676648060776e-05,
      "loss": 0.1015,
      "step": 3168,
      "task_loss": 0.029302000999450684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7757109239248183,
      "compression_loss": 0.0,
      "distillation_loss": 0.31011509895324707,
      "epoch": 3.01,
      "learning_rate": 3.049028202295494e-05,
      "loss": 0.3013,
      "step": 3169,
      "task_loss": 0.22210848331451416
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7757247849152058,
      "compression_loss": 0.0,
      "distillation_loss": 0.24750226736068726,
      "epoch": 3.01,
      "learning_rate": 3.0479886400512937e-05,
      "loss": 0.2541,
      "step": 3170,
      "task_loss": 0.31305140256881714
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.775738640631244,
      "compression_loss": 0.0,
      "distillation_loss": 0.3639996349811554,
      "epoch": 3.01,
      "learning_rate": 3.0469489782623163e-05,
      "loss": 0.3661,
      "step": 3171,
      "task_loss": 0.3845079243183136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7757524910739364,
      "compression_loss": 0.0,
      "distillation_loss": 0.20695006847381592,
      "epoch": 3.01,
      "learning_rate": 3.045909217117424e-05,
      "loss": 0.2035,
      "step": 3172,
      "task_loss": 0.1726771593093872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7757663362442868,
      "compression_loss": 0.0,
      "distillation_loss": 0.511163592338562,
      "epoch": 3.01,
      "learning_rate": 3.0448693568054924e-05,
      "loss": 0.5024,
      "step": 3173,
      "task_loss": 0.42367231845855713
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7757801761432989,
      "compression_loss": 0.0,
      "distillation_loss": 0.4330875873565674,
      "epoch": 3.01,
      "learning_rate": 3.0438293975154186e-05,
      "loss": 0.4055,
      "step": 3174,
      "task_loss": 0.157136470079422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7757940107719762,
      "compression_loss": 0.0,
      "distillation_loss": 0.3330579698085785,
      "epoch": 3.02,
      "learning_rate": 3.042789339436116e-05,
      "loss": 0.3187,
      "step": 3175,
      "task_loss": 0.18938302993774414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7758078401313225,
      "compression_loss": 0.0,
      "distillation_loss": 0.26333099603652954,
      "epoch": 3.02,
      "learning_rate": 3.041749182756515e-05,
      "loss": 0.2615,
      "step": 3176,
      "task_loss": 0.24524812400341034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7758216642223416,
      "compression_loss": 0.0,
      "distillation_loss": 0.11996375769376755,
      "epoch": 3.02,
      "learning_rate": 3.0407089276655664e-05,
      "loss": 0.1205,
      "step": 3177,
      "task_loss": 0.12564218044281006
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.775835483046037,
      "compression_loss": 0.0,
      "distillation_loss": 0.18018177151679993,
      "epoch": 3.02,
      "learning_rate": 3.039668574352237e-05,
      "loss": 0.1752,
      "step": 3178,
      "task_loss": 0.12995178997516632
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7758492966034125,
      "compression_loss": 0.0,
      "distillation_loss": 0.23442718386650085,
      "epoch": 3.02,
      "learning_rate": 3.0386281230055113e-05,
      "loss": 0.2305,
      "step": 3179,
      "task_loss": 0.19505998492240906
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7758631048954718,
      "compression_loss": 0.0,
      "distillation_loss": 0.22563649713993073,
      "epoch": 3.02,
      "learning_rate": 3.0375875738143938e-05,
      "loss": 0.2198,
      "step": 3180,
      "task_loss": 0.16684770584106445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7758769079232184,
      "compression_loss": 0.0,
      "distillation_loss": 0.05871054530143738,
      "epoch": 3.02,
      "learning_rate": 3.0365469269679042e-05,
      "loss": 0.0537,
      "step": 3181,
      "task_loss": 0.008890055119991302
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7758907056876562,
      "compression_loss": 0.0,
      "distillation_loss": 0.07384837418794632,
      "epoch": 3.02,
      "learning_rate": 3.0355061826550813e-05,
      "loss": 0.078,
      "step": 3182,
      "task_loss": 0.1150825098156929
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7759044981897888,
      "compression_loss": 0.0,
      "distillation_loss": 0.29694080352783203,
      "epoch": 3.02,
      "learning_rate": 3.0344653410649815e-05,
      "loss": 0.2839,
      "step": 3183,
      "task_loss": 0.1665906310081482
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7759182854306198,
      "compression_loss": 0.0,
      "distillation_loss": 0.06391322612762451,
      "epoch": 3.02,
      "learning_rate": 3.033424402386678e-05,
      "loss": 0.0754,
      "step": 3184,
      "task_loss": 0.1787717342376709
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.775932067411153,
      "compression_loss": 0.0,
      "distillation_loss": 0.09081655740737915,
      "epoch": 3.02,
      "learning_rate": 3.032383366809263e-05,
      "loss": 0.0906,
      "step": 3185,
      "task_loss": 0.088164322078228
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.775945844132392,
      "compression_loss": 0.0,
      "distillation_loss": 0.2625811696052551,
      "epoch": 3.03,
      "learning_rate": 3.031342234521845e-05,
      "loss": 0.2886,
      "step": 3186,
      "task_loss": 0.5232532620429993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7759596155953404,
      "compression_loss": 0.0,
      "distillation_loss": 0.05291663110256195,
      "epoch": 3.03,
      "learning_rate": 3.030301005713552e-05,
      "loss": 0.0481,
      "step": 3187,
      "task_loss": 0.004770837724208832
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7759733818010022,
      "compression_loss": 0.0,
      "distillation_loss": 0.15582989156246185,
      "epoch": 3.03,
      "learning_rate": 3.0292596805735274e-05,
      "loss": 0.1483,
      "step": 3188,
      "task_loss": 0.08084064722061157
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7759871427503807,
      "compression_loss": 0.0,
      "distillation_loss": 0.20074017345905304,
      "epoch": 3.03,
      "learning_rate": 3.028218259290932e-05,
      "loss": 0.1874,
      "step": 3189,
      "task_loss": 0.06705089658498764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.77600089844448,
      "compression_loss": 0.0,
      "distillation_loss": 0.06447502970695496,
      "epoch": 3.03,
      "learning_rate": 3.0271767420549463e-05,
      "loss": 0.0681,
      "step": 3190,
      "task_loss": 0.1006789430975914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7760146488843033,
      "compression_loss": 0.0,
      "distillation_loss": 0.2723419964313507,
      "epoch": 3.03,
      "learning_rate": 3.0261351290547667e-05,
      "loss": 0.2704,
      "step": 3191,
      "task_loss": 0.25289347767829895
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7760283940708547,
      "compression_loss": 0.0,
      "distillation_loss": 0.2833019196987152,
      "epoch": 3.03,
      "learning_rate": 3.025093420479607e-05,
      "loss": 0.2691,
      "step": 3192,
      "task_loss": 0.1415441334247589
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7760421340051376,
      "compression_loss": 0.0,
      "distillation_loss": 0.12315277010202408,
      "epoch": 3.03,
      "learning_rate": 3.0240516165186976e-05,
      "loss": 0.1165,
      "step": 3193,
      "task_loss": 0.056251607835292816
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7760558686881558,
      "compression_loss": 0.0,
      "distillation_loss": 0.39456677436828613,
      "epoch": 3.03,
      "learning_rate": 3.0230097173612896e-05,
      "loss": 0.3795,
      "step": 3194,
      "task_loss": 0.24376647174358368
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.776069598120913,
      "compression_loss": 0.0,
      "distillation_loss": 0.13045181334018707,
      "epoch": 3.03,
      "learning_rate": 3.021967723196647e-05,
      "loss": 0.1243,
      "step": 3195,
      "task_loss": 0.06902102380990982
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7760833223044128,
      "compression_loss": 0.0,
      "distillation_loss": 0.13044685125350952,
      "epoch": 3.04,
      "learning_rate": 3.020925634214054e-05,
      "loss": 0.1433,
      "step": 3196,
      "task_loss": 0.25945261120796204
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.776097041239659,
      "compression_loss": 0.0,
      "distillation_loss": 0.07756803929805756,
      "epoch": 3.04,
      "learning_rate": 3.01988345060281e-05,
      "loss": 0.0819,
      "step": 3197,
      "task_loss": 0.12053569406270981
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7761107549276551,
      "compression_loss": 0.0,
      "distillation_loss": 0.285247802734375,
      "epoch": 3.04,
      "learning_rate": 3.018841172552234e-05,
      "loss": 0.2711,
      "step": 3198,
      "task_loss": 0.14363518357276917
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7761244633694051,
      "compression_loss": 0.0,
      "distillation_loss": 0.36870819330215454,
      "epoch": 3.04,
      "learning_rate": 3.01779880025166e-05,
      "loss": 0.3675,
      "step": 3199,
      "task_loss": 0.3562384843826294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7761381665659124,
      "compression_loss": 0.0,
      "distillation_loss": 0.09676843881607056,
      "epoch": 3.04,
      "learning_rate": 3.0167563338904402e-05,
      "loss": 0.0949,
      "step": 3200,
      "task_loss": 0.0785759687423706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7761518645181807,
      "compression_loss": 0.0,
      "distillation_loss": 0.0968627855181694,
      "epoch": 3.04,
      "learning_rate": 3.0157137736579445e-05,
      "loss": 0.0951,
      "step": 3201,
      "task_loss": 0.07877330482006073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7761655572272138,
      "compression_loss": 0.0,
      "distillation_loss": 0.2662609815597534,
      "epoch": 3.04,
      "learning_rate": 3.014671119743556e-05,
      "loss": 0.2659,
      "step": 3202,
      "task_loss": 0.26285213232040405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7761792446940153,
      "compression_loss": 0.0,
      "distillation_loss": 0.07620637863874435,
      "epoch": 3.04,
      "learning_rate": 3.013628372336682e-05,
      "loss": 0.0915,
      "step": 3203,
      "task_loss": 0.22959944605827332
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.776192926919589,
      "compression_loss": 0.0,
      "distillation_loss": 0.13465449213981628,
      "epoch": 3.04,
      "learning_rate": 3.0125855316267394e-05,
      "loss": 0.1472,
      "step": 3204,
      "task_loss": 0.2603115439414978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7762066039049383,
      "compression_loss": 0.0,
      "distillation_loss": 0.14497964084148407,
      "epoch": 3.04,
      "learning_rate": 3.0115425978031663e-05,
      "loss": 0.1501,
      "step": 3205,
      "task_loss": 0.1961698830127716
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7762202756510672,
      "compression_loss": 0.0,
      "distillation_loss": 0.24775618314743042,
      "epoch": 3.04,
      "learning_rate": 3.0104995710554174e-05,
      "loss": 0.2393,
      "step": 3206,
      "task_loss": 0.16288542747497559
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7762339421589793,
      "compression_loss": 0.0,
      "distillation_loss": 0.13831278681755066,
      "epoch": 3.05,
      "learning_rate": 3.0094564515729623e-05,
      "loss": 0.1352,
      "step": 3207,
      "task_loss": 0.10688167065382004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7762476034296781,
      "compression_loss": 0.0,
      "distillation_loss": 0.04619987681508064,
      "epoch": 3.05,
      "learning_rate": 3.0084132395452896e-05,
      "loss": 0.0428,
      "step": 3208,
      "task_loss": 0.011970948427915573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7762612594641676,
      "compression_loss": 0.0,
      "distillation_loss": 0.048860594630241394,
      "epoch": 3.05,
      "learning_rate": 3.0073699351619033e-05,
      "loss": 0.0447,
      "step": 3209,
      "task_loss": 0.0075436122715473175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7762749102634512,
      "compression_loss": 0.0,
      "distillation_loss": 0.09679286926984787,
      "epoch": 3.05,
      "learning_rate": 3.0063265386123247e-05,
      "loss": 0.0926,
      "step": 3210,
      "task_loss": 0.054950546473264694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7762885558285327,
      "compression_loss": 0.0,
      "distillation_loss": 0.47562968730926514,
      "epoch": 3.05,
      "learning_rate": 3.0052830500860912e-05,
      "loss": 0.4544,
      "step": 3211,
      "task_loss": 0.26370471715927124
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7763021961604157,
      "compression_loss": 0.0,
      "distillation_loss": 0.10135069489479065,
      "epoch": 3.05,
      "learning_rate": 3.0042394697727587e-05,
      "loss": 0.0972,
      "step": 3212,
      "task_loss": 0.059726666659116745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7763158312601041,
      "compression_loss": 0.0,
      "distillation_loss": 0.17847372591495514,
      "epoch": 3.05,
      "learning_rate": 3.0031957978618986e-05,
      "loss": 0.1807,
      "step": 3213,
      "task_loss": 0.20080722868442535
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7763294611286014,
      "compression_loss": 0.0,
      "distillation_loss": 0.17386582493782043,
      "epoch": 3.05,
      "learning_rate": 3.002152034543098e-05,
      "loss": 0.1664,
      "step": 3214,
      "task_loss": 0.09940002113580704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7763430857669111,
      "compression_loss": 0.0,
      "distillation_loss": 0.10909870266914368,
      "epoch": 3.05,
      "learning_rate": 3.0011081800059616e-05,
      "loss": 0.1128,
      "step": 3215,
      "task_loss": 0.14585895836353302
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7763567051760373,
      "compression_loss": 0.0,
      "distillation_loss": 0.045543644577264786,
      "epoch": 3.05,
      "learning_rate": 3.0000642344401113e-05,
      "loss": 0.0494,
      "step": 3216,
      "task_loss": 0.08457530289888382
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7763703193569834,
      "compression_loss": 0.0,
      "distillation_loss": 0.14089834690093994,
      "epoch": 3.06,
      "learning_rate": 2.999020198035184e-05,
      "loss": 0.1429,
      "step": 3217,
      "task_loss": 0.16091729700565338
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7763839283107532,
      "compression_loss": 0.0,
      "distillation_loss": 0.15442857146263123,
      "epoch": 3.06,
      "learning_rate": 2.997976070980836e-05,
      "loss": 0.1538,
      "step": 3218,
      "task_loss": 0.1483321487903595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7763975320383504,
      "compression_loss": 0.0,
      "distillation_loss": 0.1512613296508789,
      "epoch": 3.06,
      "learning_rate": 2.996931853466734e-05,
      "loss": 0.1643,
      "step": 3219,
      "task_loss": 0.28167420625686646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7764111305407785,
      "compression_loss": 0.0,
      "distillation_loss": 0.129611536860466,
      "epoch": 3.06,
      "learning_rate": 2.9958875456825692e-05,
      "loss": 0.1331,
      "step": 3220,
      "task_loss": 0.1641448736190796
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7764247238190414,
      "compression_loss": 0.0,
      "distillation_loss": 0.06467771530151367,
      "epoch": 3.06,
      "learning_rate": 2.9948431478180434e-05,
      "loss": 0.0646,
      "step": 3221,
      "task_loss": 0.06406955420970917
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7764383118741426,
      "compression_loss": 0.0,
      "distillation_loss": 0.040332306176424026,
      "epoch": 3.06,
      "learning_rate": 2.9937986600628758e-05,
      "loss": 0.0371,
      "step": 3222,
      "task_loss": 0.008311469107866287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.776451894707086,
      "compression_loss": 0.0,
      "distillation_loss": 0.23755374550819397,
      "epoch": 3.06,
      "learning_rate": 2.992754082606804e-05,
      "loss": 0.2371,
      "step": 3223,
      "task_loss": 0.23277811706066132
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.776465472318875,
      "compression_loss": 0.0,
      "distillation_loss": 0.2364281862974167,
      "epoch": 3.06,
      "learning_rate": 2.9917094156395796e-05,
      "loss": 0.2324,
      "step": 3224,
      "task_loss": 0.19599126279354095
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7764790447105135,
      "compression_loss": 0.0,
      "distillation_loss": 0.058149438351392746,
      "epoch": 3.06,
      "learning_rate": 2.990664659350973e-05,
      "loss": 0.0539,
      "step": 3225,
      "task_loss": 0.015191374346613884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7764926118830051,
      "compression_loss": 0.0,
      "distillation_loss": 0.20538832247257233,
      "epoch": 3.06,
      "learning_rate": 2.9896198139307668e-05,
      "loss": 0.1981,
      "step": 3226,
      "task_loss": 0.13257679343223572
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7765061738373535,
      "compression_loss": 0.0,
      "distillation_loss": 0.033098138868808746,
      "epoch": 3.06,
      "learning_rate": 2.9885748795687642e-05,
      "loss": 0.0396,
      "step": 3227,
      "task_loss": 0.09805357456207275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7765197305745625,
      "compression_loss": 0.0,
      "distillation_loss": 0.13397681713104248,
      "epoch": 3.07,
      "learning_rate": 2.9875298564547805e-05,
      "loss": 0.1336,
      "step": 3228,
      "task_loss": 0.1301775723695755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7765332820956355,
      "compression_loss": 0.0,
      "distillation_loss": 0.1260242760181427,
      "epoch": 3.07,
      "learning_rate": 2.9864847447786503e-05,
      "loss": 0.1186,
      "step": 3229,
      "task_loss": 0.05128313973546028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7765468284015764,
      "compression_loss": 0.0,
      "distillation_loss": 0.14219728112220764,
      "epoch": 3.07,
      "learning_rate": 2.9854395447302246e-05,
      "loss": 0.1386,
      "step": 3230,
      "task_loss": 0.10616907477378845
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7765603694933888,
      "compression_loss": 0.0,
      "distillation_loss": 0.06799346953630447,
      "epoch": 3.07,
      "learning_rate": 2.9843942564993672e-05,
      "loss": 0.0646,
      "step": 3231,
      "task_loss": 0.0343768373131752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7765739053720765,
      "compression_loss": 0.0,
      "distillation_loss": 0.03196606785058975,
      "epoch": 3.07,
      "learning_rate": 2.98334888027596e-05,
      "loss": 0.0436,
      "step": 3232,
      "task_loss": 0.14865921437740326
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.776587436038643,
      "compression_loss": 0.0,
      "distillation_loss": 0.12633199989795685,
      "epoch": 3.07,
      "learning_rate": 2.9823034162499007e-05,
      "loss": 0.1216,
      "step": 3233,
      "task_loss": 0.07861751317977905
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7766009614940921,
      "compression_loss": 0.0,
      "distillation_loss": 0.11836487054824829,
      "epoch": 3.07,
      "learning_rate": 2.981257864611104e-05,
      "loss": 0.1123,
      "step": 3234,
      "task_loss": 0.058093033730983734
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7766144817394275,
      "compression_loss": 0.0,
      "distillation_loss": 0.09729697555303574,
      "epoch": 3.07,
      "learning_rate": 2.980212225549498e-05,
      "loss": 0.0916,
      "step": 3235,
      "task_loss": 0.040597811341285706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7766279967756528,
      "compression_loss": 0.0,
      "distillation_loss": 0.043897949159145355,
      "epoch": 3.07,
      "learning_rate": 2.9791664992550273e-05,
      "loss": 0.0408,
      "step": 3236,
      "task_loss": 0.012871745973825455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7766415066037717,
      "compression_loss": 0.0,
      "distillation_loss": 0.27021345496177673,
      "epoch": 3.07,
      "learning_rate": 2.978120685917656e-05,
      "loss": 0.2581,
      "step": 3237,
      "task_loss": 0.1487370729446411
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7766550112247879,
      "compression_loss": 0.0,
      "distillation_loss": 0.24136856198310852,
      "epoch": 3.08,
      "learning_rate": 2.9770747857273584e-05,
      "loss": 0.2404,
      "step": 3238,
      "task_loss": 0.23204952478408813
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7766685106397051,
      "compression_loss": 0.0,
      "distillation_loss": 0.038442693650722504,
      "epoch": 3.08,
      "learning_rate": 2.9760287988741293e-05,
      "loss": 0.0354,
      "step": 3239,
      "task_loss": 0.008406125009059906
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.776682004849527,
      "compression_loss": 0.0,
      "distillation_loss": 0.2253015786409378,
      "epoch": 3.08,
      "learning_rate": 2.9749827255479755e-05,
      "loss": 0.2234,
      "step": 3240,
      "task_loss": 0.20639631152153015
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7766954938552572,
      "compression_loss": 0.0,
      "distillation_loss": 0.19598889350891113,
      "epoch": 3.08,
      "learning_rate": 2.9739365659389223e-05,
      "loss": 0.1854,
      "step": 3241,
      "task_loss": 0.08999612182378769
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7767089776578995,
      "compression_loss": 0.0,
      "distillation_loss": 0.029667746275663376,
      "epoch": 3.08,
      "learning_rate": 2.972890320237009e-05,
      "loss": 0.0271,
      "step": 3242,
      "task_loss": 0.0044384244829416275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7767224562584574,
      "compression_loss": 0.0,
      "distillation_loss": 0.03521028161048889,
      "epoch": 3.08,
      "learning_rate": 2.971843988632292e-05,
      "loss": 0.0321,
      "step": 3243,
      "task_loss": 0.00444982573390007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7767359296579347,
      "compression_loss": 0.0,
      "distillation_loss": 0.06014445796608925,
      "epoch": 3.08,
      "learning_rate": 2.970797571314842e-05,
      "loss": 0.0566,
      "step": 3244,
      "task_loss": 0.02440224587917328
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7767493978573352,
      "compression_loss": 0.0,
      "distillation_loss": 0.27860748767852783,
      "epoch": 3.08,
      "learning_rate": 2.9697510684747454e-05,
      "loss": 0.2662,
      "step": 3245,
      "task_loss": 0.15431803464889526
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7767628608576623,
      "compression_loss": 0.0,
      "distillation_loss": 0.36406850814819336,
      "epoch": 3.08,
      "learning_rate": 2.9687044803021057e-05,
      "loss": 0.3679,
      "step": 3246,
      "task_loss": 0.4021039605140686
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.77677631865992,
      "compression_loss": 0.0,
      "distillation_loss": 0.4641655385494232,
      "epoch": 3.08,
      "learning_rate": 2.9676578069870392e-05,
      "loss": 0.4604,
      "step": 3247,
      "task_loss": 0.4263751804828644
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7767897712651117,
      "compression_loss": 0.0,
      "distillation_loss": 0.10289011895656586,
      "epoch": 3.08,
      "learning_rate": 2.9666110487196798e-05,
      "loss": 0.1047,
      "step": 3248,
      "task_loss": 0.12141219526529312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7768032186742413,
      "compression_loss": 0.0,
      "distillation_loss": 0.09398964792490005,
      "epoch": 3.09,
      "learning_rate": 2.9655642056901762e-05,
      "loss": 0.0971,
      "step": 3249,
      "task_loss": 0.1248759999871254
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7768166608883122,
      "compression_loss": 0.0,
      "distillation_loss": 0.20722612738609314,
      "epoch": 3.09,
      "learning_rate": 2.9645172780886927e-05,
      "loss": 0.2046,
      "step": 3250,
      "task_loss": 0.18093663454055786
    },
    {
      "epoch": 3.09,
      "eval_accuracy": 0.8795871559633027,
      "eval_loss": 0.5020281076431274,
      "eval_runtime": 17.7871,
      "eval_samples_per_second": 49.024,
      "eval_steps_per_second": 6.128,
      "step": 3250
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7768300979083285,
      "compression_loss": 0.0,
      "distillation_loss": 0.0628579705953598,
      "epoch": 3.09,
      "learning_rate": 2.9634702661054085e-05,
      "loss": 0.0577,
      "step": 3251,
      "task_loss": 0.010829754173755646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7768435297352936,
      "compression_loss": 0.0,
      "distillation_loss": 0.04929596930742264,
      "epoch": 3.09,
      "learning_rate": 2.962423169930518e-05,
      "loss": 0.0538,
      "step": 3252,
      "task_loss": 0.09398236870765686
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7768569563702111,
      "compression_loss": 0.0,
      "distillation_loss": 0.055220600217580795,
      "epoch": 3.09,
      "learning_rate": 2.961375989754232e-05,
      "loss": 0.0594,
      "step": 3253,
      "task_loss": 0.09694081544876099
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.776870377814085,
      "compression_loss": 0.0,
      "distillation_loss": 0.14030757546424866,
      "epoch": 3.09,
      "learning_rate": 2.9603287257667754e-05,
      "loss": 0.1347,
      "step": 3254,
      "task_loss": 0.08383115381002426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7768837940679186,
      "compression_loss": 0.0,
      "distillation_loss": 0.05990925058722496,
      "epoch": 3.09,
      "learning_rate": 2.9592813781583885e-05,
      "loss": 0.0712,
      "step": 3255,
      "task_loss": 0.17235514521598816
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.776897205132716,
      "compression_loss": 0.0,
      "distillation_loss": 0.14225684106349945,
      "epoch": 3.09,
      "learning_rate": 2.958233947119328e-05,
      "loss": 0.144,
      "step": 3256,
      "task_loss": 0.16001349687576294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7769106110094804,
      "compression_loss": 0.0,
      "distillation_loss": 0.03743474930524826,
      "epoch": 3.09,
      "learning_rate": 2.9571864328398636e-05,
      "loss": 0.0344,
      "step": 3257,
      "task_loss": 0.006805334240198135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.776924011699216,
      "compression_loss": 0.0,
      "distillation_loss": 0.13843439519405365,
      "epoch": 3.09,
      "learning_rate": 2.956138835510282e-05,
      "loss": 0.1333,
      "step": 3258,
      "task_loss": 0.08740270882844925
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.776937407202926,
      "compression_loss": 0.0,
      "distillation_loss": 0.044483255594968796,
      "epoch": 3.09,
      "learning_rate": 2.9550911553208838e-05,
      "loss": 0.0404,
      "step": 3259,
      "task_loss": 0.003829212859272957
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7769507975216144,
      "compression_loss": 0.0,
      "distillation_loss": 0.3198070526123047,
      "epoch": 3.1,
      "learning_rate": 2.954043392461986e-05,
      "loss": 0.3065,
      "step": 3260,
      "task_loss": 0.18672311305999756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7769641826562849,
      "compression_loss": 0.0,
      "distillation_loss": 0.1478300392627716,
      "epoch": 3.1,
      "learning_rate": 2.952995547123919e-05,
      "loss": 0.1487,
      "step": 3261,
      "task_loss": 0.1560606062412262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7769775626079409,
      "compression_loss": 0.0,
      "distillation_loss": 0.05190206691622734,
      "epoch": 3.1,
      "learning_rate": 2.9519476194970286e-05,
      "loss": 0.0471,
      "step": 3262,
      "task_loss": 0.0037672966718673706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7769909373775864,
      "compression_loss": 0.0,
      "distillation_loss": 0.07992243766784668,
      "epoch": 3.1,
      "learning_rate": 2.9508996097716777e-05,
      "loss": 0.0892,
      "step": 3263,
      "task_loss": 0.17313553392887115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7770043069662249,
      "compression_loss": 0.0,
      "distillation_loss": 0.1712629646062851,
      "epoch": 3.1,
      "learning_rate": 2.949851518138241e-05,
      "loss": 0.1902,
      "step": 3264,
      "task_loss": 0.3601612150669098
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7770176713748601,
      "compression_loss": 0.0,
      "distillation_loss": 0.020557358860969543,
      "epoch": 3.1,
      "learning_rate": 2.948803344787109e-05,
      "loss": 0.019,
      "step": 3265,
      "task_loss": 0.004615806043148041
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7770310306044956,
      "compression_loss": 0.0,
      "distillation_loss": 0.058595433831214905,
      "epoch": 3.1,
      "learning_rate": 2.947755089908688e-05,
      "loss": 0.0613,
      "step": 3266,
      "task_loss": 0.0856269896030426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7770443846561353,
      "compression_loss": 0.0,
      "distillation_loss": 0.13503427803516388,
      "epoch": 3.1,
      "learning_rate": 2.946706753693398e-05,
      "loss": 0.1376,
      "step": 3267,
      "task_loss": 0.16037359833717346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7770577335307828,
      "compression_loss": 0.0,
      "distillation_loss": 0.06786477565765381,
      "epoch": 3.1,
      "learning_rate": 2.945658336331676e-05,
      "loss": 0.0684,
      "step": 3268,
      "task_loss": 0.07298760861158371
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7770710772294416,
      "compression_loss": 0.0,
      "distillation_loss": 0.06129030883312225,
      "epoch": 3.1,
      "learning_rate": 2.9446098380139703e-05,
      "loss": 0.0586,
      "step": 3269,
      "task_loss": 0.03390103578567505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7770844157531156,
      "compression_loss": 0.0,
      "distillation_loss": 0.07370973378419876,
      "epoch": 3.11,
      "learning_rate": 2.9435612589307458e-05,
      "loss": 0.0731,
      "step": 3270,
      "task_loss": 0.06739000231027603
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7770977491028085,
      "compression_loss": 0.0,
      "distillation_loss": 0.0707666426897049,
      "epoch": 3.11,
      "learning_rate": 2.942512599272483e-05,
      "loss": 0.0739,
      "step": 3271,
      "task_loss": 0.1019490510225296
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7771110772795238,
      "compression_loss": 0.0,
      "distillation_loss": 0.08299912512302399,
      "epoch": 3.11,
      "learning_rate": 2.9414638592296752e-05,
      "loss": 0.0816,
      "step": 3272,
      "task_loss": 0.06940528750419617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7771244002842652,
      "compression_loss": 0.0,
      "distillation_loss": 0.13758215308189392,
      "epoch": 3.11,
      "learning_rate": 2.9404150389928316e-05,
      "loss": 0.1317,
      "step": 3273,
      "task_loss": 0.07901225984096527
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7771377181180366,
      "compression_loss": 0.0,
      "distillation_loss": 0.14945004880428314,
      "epoch": 3.11,
      "learning_rate": 2.9393661387524745e-05,
      "loss": 0.1439,
      "step": 3274,
      "task_loss": 0.09377557784318924
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7771510307818414,
      "compression_loss": 0.0,
      "distillation_loss": 0.13088908791542053,
      "epoch": 3.11,
      "learning_rate": 2.9383171586991424e-05,
      "loss": 0.1261,
      "step": 3275,
      "task_loss": 0.0828273668885231
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7771643382766835,
      "compression_loss": 0.0,
      "distillation_loss": 0.1356673389673233,
      "epoch": 3.11,
      "learning_rate": 2.9372680990233875e-05,
      "loss": 0.1301,
      "step": 3276,
      "task_loss": 0.07979649305343628
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7771776406035665,
      "compression_loss": 0.0,
      "distillation_loss": 0.15914839506149292,
      "epoch": 3.11,
      "learning_rate": 2.9362189599157776e-05,
      "loss": 0.1522,
      "step": 3277,
      "task_loss": 0.08918256312608719
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7771909377634941,
      "compression_loss": 0.0,
      "distillation_loss": 0.031945958733558655,
      "epoch": 3.11,
      "learning_rate": 2.9351697415668917e-05,
      "loss": 0.0375,
      "step": 3278,
      "task_loss": 0.08777586370706558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7772042297574699,
      "compression_loss": 0.0,
      "distillation_loss": 0.07947254180908203,
      "epoch": 3.11,
      "learning_rate": 2.9341204441673266e-05,
      "loss": 0.0822,
      "step": 3279,
      "task_loss": 0.106949083507061
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7772175165864978,
      "compression_loss": 0.0,
      "distillation_loss": 0.11136071383953094,
      "epoch": 3.11,
      "learning_rate": 2.9330710679076916e-05,
      "loss": 0.1059,
      "step": 3280,
      "task_loss": 0.057253483682870865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7772307982515813,
      "compression_loss": 0.0,
      "distillation_loss": 0.04170932248234749,
      "epoch": 3.12,
      "learning_rate": 2.9320216129786116e-05,
      "loss": 0.0457,
      "step": 3281,
      "task_loss": 0.08145752549171448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.777244074753724,
      "compression_loss": 0.0,
      "distillation_loss": 0.014708688482642174,
      "epoch": 3.12,
      "learning_rate": 2.9309720795707257e-05,
      "loss": 0.0136,
      "step": 3282,
      "task_loss": 0.0032885856926441193
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7772573460939297,
      "compression_loss": 0.0,
      "distillation_loss": 0.08308391273021698,
      "epoch": 3.12,
      "learning_rate": 2.9299224678746855e-05,
      "loss": 0.0762,
      "step": 3283,
      "task_loss": 0.014657005667686462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7772706122732022,
      "compression_loss": 0.0,
      "distillation_loss": 0.1468481719493866,
      "epoch": 3.12,
      "learning_rate": 2.928872778081158e-05,
      "loss": 0.1414,
      "step": 3284,
      "task_loss": 0.09249945729970932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7772838732925449,
      "compression_loss": 0.0,
      "distillation_loss": 0.029921449720859528,
      "epoch": 3.12,
      "learning_rate": 2.9278230103808257e-05,
      "loss": 0.0345,
      "step": 3285,
      "task_loss": 0.07549962401390076
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7772971291529618,
      "compression_loss": 0.0,
      "distillation_loss": 0.0418395921587944,
      "epoch": 3.12,
      "learning_rate": 2.9267731649643827e-05,
      "loss": 0.048,
      "step": 3286,
      "task_loss": 0.10373726487159729
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7773103798554563,
      "compression_loss": 0.0,
      "distillation_loss": 0.05867641791701317,
      "epoch": 3.12,
      "learning_rate": 2.9257232420225394e-05,
      "loss": 0.0721,
      "step": 3287,
      "task_loss": 0.19298899173736572
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7773236254010323,
      "compression_loss": 0.0,
      "distillation_loss": 0.05694715678691864,
      "epoch": 3.12,
      "learning_rate": 2.9246732417460178e-05,
      "loss": 0.0553,
      "step": 3288,
      "task_loss": 0.0401817262172699
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7773368657906933,
      "compression_loss": 0.0,
      "distillation_loss": 0.056060872972011566,
      "epoch": 3.12,
      "learning_rate": 2.9236231643255578e-05,
      "loss": 0.0606,
      "step": 3289,
      "task_loss": 0.10119974613189697
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7773501010254431,
      "compression_loss": 0.0,
      "distillation_loss": 0.08849294483661652,
      "epoch": 3.12,
      "learning_rate": 2.922573009951909e-05,
      "loss": 0.0865,
      "step": 3290,
      "task_loss": 0.06868458539247513
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7773633311062854,
      "compression_loss": 0.0,
      "distillation_loss": 0.0786486566066742,
      "epoch": 3.13,
      "learning_rate": 2.9215227788158382e-05,
      "loss": 0.0987,
      "step": 3291,
      "task_loss": 0.27886343002319336
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7773765560342237,
      "compression_loss": 0.0,
      "distillation_loss": 0.08726784586906433,
      "epoch": 3.13,
      "learning_rate": 2.920472471108125e-05,
      "loss": 0.0813,
      "step": 3292,
      "task_loss": 0.0275074802339077
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.777389775810262,
      "compression_loss": 0.0,
      "distillation_loss": 0.10106363892555237,
      "epoch": 3.13,
      "learning_rate": 2.919422087019561e-05,
      "loss": 0.0965,
      "step": 3293,
      "task_loss": 0.055741891264915466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7774029904354037,
      "compression_loss": 0.0,
      "distillation_loss": 0.14217272400856018,
      "epoch": 3.13,
      "learning_rate": 2.9183716267409562e-05,
      "loss": 0.1446,
      "step": 3294,
      "task_loss": 0.16670069098472595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7774161999106526,
      "compression_loss": 0.0,
      "distillation_loss": 0.10469906777143478,
      "epoch": 3.13,
      "learning_rate": 2.9173210904631297e-05,
      "loss": 0.1019,
      "step": 3295,
      "task_loss": 0.07687127590179443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7774294042370123,
      "compression_loss": 0.0,
      "distillation_loss": 0.17787286639213562,
      "epoch": 3.13,
      "learning_rate": 2.916270478376918e-05,
      "loss": 0.1797,
      "step": 3296,
      "task_loss": 0.19574972987174988
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7774426034154867,
      "compression_loss": 0.0,
      "distillation_loss": 0.22817805409431458,
      "epoch": 3.13,
      "learning_rate": 2.9152197906731687e-05,
      "loss": 0.2223,
      "step": 3297,
      "task_loss": 0.16891571879386902
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7774557974470792,
      "compression_loss": 0.0,
      "distillation_loss": 0.2010933756828308,
      "epoch": 3.13,
      "learning_rate": 2.9141690275427445e-05,
      "loss": 0.2089,
      "step": 3298,
      "task_loss": 0.27871859073638916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7774689863327936,
      "compression_loss": 0.0,
      "distillation_loss": 0.07105930149555206,
      "epoch": 3.13,
      "learning_rate": 2.9131181891765226e-05,
      "loss": 0.067,
      "step": 3299,
      "task_loss": 0.030665744096040726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7774821700736336,
      "compression_loss": 0.0,
      "distillation_loss": 0.03237374126911163,
      "epoch": 3.13,
      "learning_rate": 2.9120672757653916e-05,
      "loss": 0.0299,
      "step": 3300,
      "task_loss": 0.00755789689719677
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.777495348670603,
      "compression_loss": 0.0,
      "distillation_loss": 0.04193927347660065,
      "epoch": 3.13,
      "learning_rate": 2.9110162875002552e-05,
      "loss": 0.0383,
      "step": 3301,
      "task_loss": 0.005132727324962616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7775085221247052,
      "compression_loss": 0.0,
      "distillation_loss": 0.02390095964074135,
      "epoch": 3.14,
      "learning_rate": 2.909965224572031e-05,
      "loss": 0.0227,
      "step": 3302,
      "task_loss": 0.011394314467906952
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7775216904369441,
      "compression_loss": 0.0,
      "distillation_loss": 0.11238003522157669,
      "epoch": 3.14,
      "learning_rate": 2.9089140871716492e-05,
      "loss": 0.1184,
      "step": 3303,
      "task_loss": 0.1725841611623764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7775348536083233,
      "compression_loss": 0.0,
      "distillation_loss": 0.09369160979986191,
      "epoch": 3.14,
      "learning_rate": 2.9078628754900543e-05,
      "loss": 0.0978,
      "step": 3304,
      "task_loss": 0.1350669115781784
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7775480116398465,
      "compression_loss": 0.0,
      "distillation_loss": 0.04276084154844284,
      "epoch": 3.14,
      "learning_rate": 2.9068115897182036e-05,
      "loss": 0.0476,
      "step": 3305,
      "task_loss": 0.09119967371225357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7775611645325174,
      "compression_loss": 0.0,
      "distillation_loss": 0.14125396311283112,
      "epoch": 3.14,
      "learning_rate": 2.905760230047068e-05,
      "loss": 0.1457,
      "step": 3306,
      "task_loss": 0.18620248138904572
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7775743122873398,
      "compression_loss": 0.0,
      "distillation_loss": 0.05440014600753784,
      "epoch": 3.14,
      "learning_rate": 2.9047087966676327e-05,
      "loss": 0.0512,
      "step": 3307,
      "task_loss": 0.021976150572299957
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.777587454905317,
      "compression_loss": 0.0,
      "distillation_loss": 0.10159926116466522,
      "epoch": 3.14,
      "learning_rate": 2.903657289770896e-05,
      "loss": 0.0933,
      "step": 3308,
      "task_loss": 0.018225595355033875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.777600592387453,
      "compression_loss": 0.0,
      "distillation_loss": 0.05529090762138367,
      "epoch": 3.14,
      "learning_rate": 2.902605709547868e-05,
      "loss": 0.0516,
      "step": 3309,
      "task_loss": 0.01865418255329132
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7776137247347515,
      "compression_loss": 0.0,
      "distillation_loss": 0.15098199248313904,
      "epoch": 3.14,
      "learning_rate": 2.9015540561895738e-05,
      "loss": 0.1434,
      "step": 3310,
      "task_loss": 0.07475800067186356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.777626851948216,
      "compression_loss": 0.0,
      "distillation_loss": 0.0264451764523983,
      "epoch": 3.14,
      "learning_rate": 2.9005023298870514e-05,
      "loss": 0.0323,
      "step": 3311,
      "task_loss": 0.08548250794410706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7776399740288504,
      "compression_loss": 0.0,
      "distillation_loss": 0.24020592868328094,
      "epoch": 3.15,
      "learning_rate": 2.8994505308313523e-05,
      "loss": 0.2399,
      "step": 3312,
      "task_loss": 0.2373048961162567
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7776530909776581,
      "compression_loss": 0.0,
      "distillation_loss": 0.16168448328971863,
      "epoch": 3.15,
      "learning_rate": 2.8983986592135404e-05,
      "loss": 0.1736,
      "step": 3313,
      "task_loss": 0.2803611159324646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7776662027956431,
      "compression_loss": 0.0,
      "distillation_loss": 0.2303021252155304,
      "epoch": 3.15,
      "learning_rate": 2.897346715224693e-05,
      "loss": 0.23,
      "step": 3314,
      "task_loss": 0.22724005579948425
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7776793094838088,
      "compression_loss": 0.0,
      "distillation_loss": 0.12179985642433167,
      "epoch": 3.15,
      "learning_rate": 2.8962946990559013e-05,
      "loss": 0.1226,
      "step": 3315,
      "task_loss": 0.1297818422317505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7776924110431591,
      "compression_loss": 0.0,
      "distillation_loss": 0.09324241429567337,
      "epoch": 3.15,
      "learning_rate": 2.8952426108982693e-05,
      "loss": 0.0956,
      "step": 3316,
      "task_loss": 0.11695347726345062
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7777055074746976,
      "compression_loss": 0.0,
      "distillation_loss": 0.06409186124801636,
      "epoch": 3.15,
      "learning_rate": 2.8941904509429134e-05,
      "loss": 0.0587,
      "step": 3317,
      "task_loss": 0.010216565802693367
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7777185987794278,
      "compression_loss": 0.0,
      "distillation_loss": 0.1214500218629837,
      "epoch": 3.15,
      "learning_rate": 2.8931382193809635e-05,
      "loss": 0.1167,
      "step": 3318,
      "task_loss": 0.0738031342625618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7777316849583537,
      "compression_loss": 0.0,
      "distillation_loss": 0.026341602206230164,
      "epoch": 3.15,
      "learning_rate": 2.8920859164035625e-05,
      "loss": 0.024,
      "step": 3319,
      "task_loss": 0.00342458114027977
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7777447660124788,
      "compression_loss": 0.0,
      "distillation_loss": 0.09703463315963745,
      "epoch": 3.15,
      "learning_rate": 2.8910335422018664e-05,
      "loss": 0.0881,
      "step": 3320,
      "task_loss": 0.00742473267018795
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7777578419428068,
      "compression_loss": 0.0,
      "distillation_loss": 0.16258884966373444,
      "epoch": 3.15,
      "learning_rate": 2.8899810969670448e-05,
      "loss": 0.1614,
      "step": 3321,
      "task_loss": 0.15071289241313934
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7777709127503414,
      "compression_loss": 0.0,
      "distillation_loss": 0.04823547974228859,
      "epoch": 3.15,
      "learning_rate": 2.8889285808902784e-05,
      "loss": 0.0598,
      "step": 3322,
      "task_loss": 0.16371826827526093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7777839784360863,
      "compression_loss": 0.0,
      "distillation_loss": 0.023103870451450348,
      "epoch": 3.16,
      "learning_rate": 2.887875994162762e-05,
      "loss": 0.0309,
      "step": 3323,
      "task_loss": 0.10059693455696106
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7777970390010451,
      "compression_loss": 0.0,
      "distillation_loss": 0.1937963366508484,
      "epoch": 3.16,
      "learning_rate": 2.886823336975703e-05,
      "loss": 0.1993,
      "step": 3324,
      "task_loss": 0.248738631606102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7778100944462216,
      "compression_loss": 0.0,
      "distillation_loss": 0.015010814182460308,
      "epoch": 3.16,
      "learning_rate": 2.885770609520323e-05,
      "loss": 0.014,
      "step": 3325,
      "task_loss": 0.004603438079357147
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7778231447726194,
      "compression_loss": 0.0,
      "distillation_loss": 0.05482349917292595,
      "epoch": 3.16,
      "learning_rate": 2.8847178119878527e-05,
      "loss": 0.0568,
      "step": 3326,
      "task_loss": 0.07430904358625412
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7778361899812423,
      "compression_loss": 0.0,
      "distillation_loss": 0.12944038212299347,
      "epoch": 3.16,
      "learning_rate": 2.883664944569539e-05,
      "loss": 0.1358,
      "step": 3327,
      "task_loss": 0.19279280304908752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7778492300730937,
      "compression_loss": 0.0,
      "distillation_loss": 0.024749279022216797,
      "epoch": 3.16,
      "learning_rate": 2.8826120074566414e-05,
      "loss": 0.0414,
      "step": 3328,
      "task_loss": 0.19170710444450378
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7778622650491777,
      "compression_loss": 0.0,
      "distillation_loss": 0.04718421772122383,
      "epoch": 3.16,
      "learning_rate": 2.8815590008404293e-05,
      "loss": 0.0442,
      "step": 3329,
      "task_loss": 0.017278321087360382
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7778752949104976,
      "compression_loss": 0.0,
      "distillation_loss": 0.06321703642606735,
      "epoch": 3.16,
      "learning_rate": 2.8805059249121874e-05,
      "loss": 0.0583,
      "step": 3330,
      "task_loss": 0.013739963993430138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7778883196580573,
      "compression_loss": 0.0,
      "distillation_loss": 0.17093795537948608,
      "epoch": 3.16,
      "learning_rate": 2.8794527798632117e-05,
      "loss": 0.1645,
      "step": 3331,
      "task_loss": 0.10624982416629791
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7779013392928603,
      "compression_loss": 0.0,
      "distillation_loss": 0.09663266688585281,
      "epoch": 3.16,
      "learning_rate": 2.8783995658848105e-05,
      "loss": 0.1007,
      "step": 3332,
      "task_loss": 0.13778991997241974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7779143538159106,
      "compression_loss": 0.0,
      "distillation_loss": 0.12523743510246277,
      "epoch": 3.17,
      "learning_rate": 2.877346283168306e-05,
      "loss": 0.1164,
      "step": 3333,
      "task_loss": 0.03714917227625847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7779273632282115,
      "compression_loss": 0.0,
      "distillation_loss": 0.1300932914018631,
      "epoch": 3.17,
      "learning_rate": 2.876292931905032e-05,
      "loss": 0.1346,
      "step": 3334,
      "task_loss": 0.17540670931339264
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7779403675307669,
      "compression_loss": 0.0,
      "distillation_loss": 0.16550663113594055,
      "epoch": 3.17,
      "learning_rate": 2.875239512286335e-05,
      "loss": 0.17,
      "step": 3335,
      "task_loss": 0.21077799797058105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7779533667245805,
      "compression_loss": 0.0,
      "distillation_loss": 0.10607212781906128,
      "epoch": 3.17,
      "learning_rate": 2.8741860245035722e-05,
      "loss": 0.0985,
      "step": 3336,
      "task_loss": 0.030357446521520615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.777966360810656,
      "compression_loss": 0.0,
      "distillation_loss": 0.0917745977640152,
      "epoch": 3.17,
      "learning_rate": 2.8731324687481176e-05,
      "loss": 0.0947,
      "step": 3337,
      "task_loss": 0.12098343670368195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7779793497899968,
      "compression_loss": 0.0,
      "distillation_loss": 0.12411828339099884,
      "epoch": 3.17,
      "learning_rate": 2.8720788452113517e-05,
      "loss": 0.115,
      "step": 3338,
      "task_loss": 0.03254596143960953
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.777992333663607,
      "compression_loss": 0.0,
      "distillation_loss": 0.12530258297920227,
      "epoch": 3.17,
      "learning_rate": 2.8710251540846723e-05,
      "loss": 0.1194,
      "step": 3339,
      "task_loss": 0.06607921421527863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7780053124324899,
      "compression_loss": 0.0,
      "distillation_loss": 0.10954394936561584,
      "epoch": 3.17,
      "learning_rate": 2.8699713955594864e-05,
      "loss": 0.1203,
      "step": 3340,
      "task_loss": 0.21743568778038025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7780182860976496,
      "compression_loss": 0.0,
      "distillation_loss": 0.047306716442108154,
      "epoch": 3.17,
      "learning_rate": 2.8689175698272147e-05,
      "loss": 0.045,
      "step": 3341,
      "task_loss": 0.02468855120241642
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7780312546600893,
      "compression_loss": 0.0,
      "distillation_loss": 0.09475719928741455,
      "epoch": 3.17,
      "learning_rate": 2.8678636770792906e-05,
      "loss": 0.0905,
      "step": 3342,
      "task_loss": 0.052515700459480286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.778044218120813,
      "compression_loss": 0.0,
      "distillation_loss": 0.1709882766008377,
      "epoch": 3.17,
      "learning_rate": 2.8668097175071572e-05,
      "loss": 0.1834,
      "step": 3343,
      "task_loss": 0.29515278339385986
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7780571764808243,
      "compression_loss": 0.0,
      "distillation_loss": 0.04090896621346474,
      "epoch": 3.18,
      "learning_rate": 2.865755691302272e-05,
      "loss": 0.0383,
      "step": 3344,
      "task_loss": 0.015103261917829514
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7780701297411269,
      "compression_loss": 0.0,
      "distillation_loss": 0.07005221396684647,
      "epoch": 3.18,
      "learning_rate": 2.864701598656104e-05,
      "loss": 0.0843,
      "step": 3345,
      "task_loss": 0.21217550337314606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7780830779027245,
      "compression_loss": 0.0,
      "distillation_loss": 0.030060699209570885,
      "epoch": 3.18,
      "learning_rate": 2.8636474397601343e-05,
      "loss": 0.0275,
      "step": 3346,
      "task_loss": 0.00450095534324646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7780960209666207,
      "compression_loss": 0.0,
      "distillation_loss": 0.2950228452682495,
      "epoch": 3.18,
      "learning_rate": 2.862593214805856e-05,
      "loss": 0.2962,
      "step": 3347,
      "task_loss": 0.30634093284606934
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7781089589338193,
      "compression_loss": 0.0,
      "distillation_loss": 0.0678018108010292,
      "epoch": 3.18,
      "learning_rate": 2.8615389239847734e-05,
      "loss": 0.0648,
      "step": 3348,
      "task_loss": 0.03745712339878082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7781218918053238,
      "compression_loss": 0.0,
      "distillation_loss": 0.1801690310239792,
      "epoch": 3.18,
      "learning_rate": 2.8604845674884045e-05,
      "loss": 0.1866,
      "step": 3349,
      "task_loss": 0.24447283148765564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.778134819582138,
      "compression_loss": 0.0,
      "distillation_loss": 0.05510277673602104,
      "epoch": 3.18,
      "learning_rate": 2.8594301455082777e-05,
      "loss": 0.0609,
      "step": 3350,
      "task_loss": 0.11345023661851883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7781477422652656,
      "compression_loss": 0.0,
      "distillation_loss": 0.1684875637292862,
      "epoch": 3.18,
      "learning_rate": 2.8583756582359338e-05,
      "loss": 0.1605,
      "step": 3351,
      "task_loss": 0.08910365402698517
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7781606598557104,
      "compression_loss": 0.0,
      "distillation_loss": 0.13622252643108368,
      "epoch": 3.18,
      "learning_rate": 2.8573211058629262e-05,
      "loss": 0.1293,
      "step": 3352,
      "task_loss": 0.06717808544635773
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7781735723544758,
      "compression_loss": 0.0,
      "distillation_loss": 0.03606577590107918,
      "epoch": 3.18,
      "learning_rate": 2.8562664885808176e-05,
      "loss": 0.0345,
      "step": 3353,
      "task_loss": 0.020131606608629227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7781864797625656,
      "compression_loss": 0.0,
      "distillation_loss": 0.07601951062679291,
      "epoch": 3.19,
      "learning_rate": 2.8552118065811868e-05,
      "loss": 0.086,
      "step": 3354,
      "task_loss": 0.17620819807052612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7781993820809836,
      "compression_loss": 0.0,
      "distillation_loss": 0.032324016094207764,
      "epoch": 3.19,
      "learning_rate": 2.85415706005562e-05,
      "loss": 0.0297,
      "step": 3355,
      "task_loss": 0.0057275425642728806
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7782122793107333,
      "compression_loss": 0.0,
      "distillation_loss": 0.08825677633285522,
      "epoch": 3.19,
      "learning_rate": 2.8531022491957178e-05,
      "loss": 0.0971,
      "step": 3356,
      "task_loss": 0.17651137709617615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7782251714528186,
      "compression_loss": 0.0,
      "distillation_loss": 0.05497463047504425,
      "epoch": 3.19,
      "learning_rate": 2.852047374193092e-05,
      "loss": 0.0514,
      "step": 3357,
      "task_loss": 0.019507795572280884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.778238058508243,
      "compression_loss": 0.0,
      "distillation_loss": 0.13246670365333557,
      "epoch": 3.19,
      "learning_rate": 2.850992435239364e-05,
      "loss": 0.1283,
      "step": 3358,
      "task_loss": 0.0911954715847969
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7782509404780101,
      "compression_loss": 0.0,
      "distillation_loss": 0.02414816804230213,
      "epoch": 3.19,
      "learning_rate": 2.8499374325261708e-05,
      "loss": 0.03,
      "step": 3359,
      "task_loss": 0.08249876648187637
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7782638173631239,
      "compression_loss": 0.0,
      "distillation_loss": 0.041819360107183456,
      "epoch": 3.19,
      "learning_rate": 2.848882366245157e-05,
      "loss": 0.0381,
      "step": 3360,
      "task_loss": 0.004362896084785461
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7782766891645878,
      "compression_loss": 0.0,
      "distillation_loss": 0.08119907230138779,
      "epoch": 3.19,
      "learning_rate": 2.847827236587982e-05,
      "loss": 0.092,
      "step": 3361,
      "task_loss": 0.18970143795013428
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7782895558834056,
      "compression_loss": 0.0,
      "distillation_loss": 0.05240718275308609,
      "epoch": 3.19,
      "learning_rate": 2.846772043746313e-05,
      "loss": 0.0514,
      "step": 3362,
      "task_loss": 0.04256965219974518
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.778302417520581,
      "compression_loss": 0.0,
      "distillation_loss": 0.1657370626926422,
      "epoch": 3.19,
      "learning_rate": 2.845716787911833e-05,
      "loss": 0.1659,
      "step": 3363,
      "task_loss": 0.16752569377422333
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7783152740771176,
      "compression_loss": 0.0,
      "distillation_loss": 0.06295058131217957,
      "epoch": 3.19,
      "learning_rate": 2.8446614692762336e-05,
      "loss": 0.0575,
      "step": 3364,
      "task_loss": 0.008071176707744598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7783281255540192,
      "compression_loss": 0.0,
      "distillation_loss": 0.147892564535141,
      "epoch": 3.2,
      "learning_rate": 2.843606088031218e-05,
      "loss": 0.1436,
      "step": 3365,
      "task_loss": 0.10463625937700272
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7783409719522894,
      "compression_loss": 0.0,
      "distillation_loss": 0.02701820805668831,
      "epoch": 3.2,
      "learning_rate": 2.842550644368502e-05,
      "loss": 0.0337,
      "step": 3366,
      "task_loss": 0.09333845973014832
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7783538132729317,
      "compression_loss": 0.0,
      "distillation_loss": 0.024246441200375557,
      "epoch": 3.2,
      "learning_rate": 2.841495138479811e-05,
      "loss": 0.0222,
      "step": 3367,
      "task_loss": 0.003893321380019188
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7783666495169501,
      "compression_loss": 0.0,
      "distillation_loss": 0.16022557020187378,
      "epoch": 3.2,
      "learning_rate": 2.8404395705568848e-05,
      "loss": 0.1523,
      "step": 3368,
      "task_loss": 0.08084793388843536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7783794806853482,
      "compression_loss": 0.0,
      "distillation_loss": 0.03452833741903305,
      "epoch": 3.2,
      "learning_rate": 2.8393839407914702e-05,
      "loss": 0.0374,
      "step": 3369,
      "task_loss": 0.06310579180717468
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7783923067791296,
      "compression_loss": 0.0,
      "distillation_loss": 0.11381056159734726,
      "epoch": 3.2,
      "learning_rate": 2.8383282493753283e-05,
      "loss": 0.1112,
      "step": 3370,
      "task_loss": 0.08721385896205902
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.778405127799298,
      "compression_loss": 0.0,
      "distillation_loss": 0.11384284496307373,
      "epoch": 3.2,
      "learning_rate": 2.83727249650023e-05,
      "loss": 0.1135,
      "step": 3371,
      "task_loss": 0.11086120456457138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7784179437468572,
      "compression_loss": 0.0,
      "distillation_loss": 0.04285869747400284,
      "epoch": 3.2,
      "learning_rate": 2.836216682357959e-05,
      "loss": 0.063,
      "step": 3372,
      "task_loss": 0.24387109279632568
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7784307546228107,
      "compression_loss": 0.0,
      "distillation_loss": 0.10957767069339752,
      "epoch": 3.2,
      "learning_rate": 2.8351608071403085e-05,
      "loss": 0.1169,
      "step": 3373,
      "task_loss": 0.18312203884124756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7784435604281623,
      "compression_loss": 0.0,
      "distillation_loss": 0.06287790834903717,
      "epoch": 3.2,
      "learning_rate": 2.8341048710390832e-05,
      "loss": 0.0677,
      "step": 3374,
      "task_loss": 0.11077691614627838
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7784563611639156,
      "compression_loss": 0.0,
      "distillation_loss": 0.035941578447818756,
      "epoch": 3.21,
      "learning_rate": 2.8330488742460987e-05,
      "loss": 0.033,
      "step": 3375,
      "task_loss": 0.006380628794431686
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7784691568310744,
      "compression_loss": 0.0,
      "distillation_loss": 0.14878231287002563,
      "epoch": 3.21,
      "learning_rate": 2.8319928169531825e-05,
      "loss": 0.1419,
      "step": 3376,
      "task_loss": 0.08023767173290253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7784819474306423,
      "compression_loss": 0.0,
      "distillation_loss": 0.15483559668064117,
      "epoch": 3.21,
      "learning_rate": 2.830936699352172e-05,
      "loss": 0.1589,
      "step": 3377,
      "task_loss": 0.1951579451560974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7784947329636229,
      "compression_loss": 0.0,
      "distillation_loss": 0.22917942702770233,
      "epoch": 3.21,
      "learning_rate": 2.8298805216349167e-05,
      "loss": 0.215,
      "step": 3378,
      "task_loss": 0.08718053251504898
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7785075134310201,
      "compression_loss": 0.0,
      "distillation_loss": 0.266732782125473,
      "epoch": 3.21,
      "learning_rate": 2.8288242839932744e-05,
      "loss": 0.2629,
      "step": 3379,
      "task_loss": 0.2285778522491455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7785202888338374,
      "compression_loss": 0.0,
      "distillation_loss": 0.09135205298662186,
      "epoch": 3.21,
      "learning_rate": 2.8277679866191194e-05,
      "loss": 0.1011,
      "step": 3380,
      "task_loss": 0.18895608186721802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7785330591730785,
      "compression_loss": 0.0,
      "distillation_loss": 0.10083561390638351,
      "epoch": 3.21,
      "learning_rate": 2.8267116297043294e-05,
      "loss": 0.1137,
      "step": 3381,
      "task_loss": 0.22926008701324463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7785458244497472,
      "compression_loss": 0.0,
      "distillation_loss": 0.020816469565033913,
      "epoch": 3.21,
      "learning_rate": 2.8256552134407993e-05,
      "loss": 0.0192,
      "step": 3382,
      "task_loss": 0.004540523514151573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7785585846648471,
      "compression_loss": 0.0,
      "distillation_loss": 0.05450872331857681,
      "epoch": 3.21,
      "learning_rate": 2.8245987380204313e-05,
      "loss": 0.0556,
      "step": 3383,
      "task_loss": 0.0653524175286293
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7785713398193818,
      "compression_loss": 0.0,
      "distillation_loss": 0.05758683755993843,
      "epoch": 3.21,
      "learning_rate": 2.8235422036351382e-05,
      "loss": 0.0611,
      "step": 3384,
      "task_loss": 0.09230072051286697
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7785840899143551,
      "compression_loss": 0.0,
      "distillation_loss": 0.1324225813150406,
      "epoch": 3.21,
      "learning_rate": 2.822485610476847e-05,
      "loss": 0.1397,
      "step": 3385,
      "task_loss": 0.2054787278175354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7785968349507707,
      "compression_loss": 0.0,
      "distillation_loss": 0.10634157061576843,
      "epoch": 3.22,
      "learning_rate": 2.8214289587374908e-05,
      "loss": 0.1103,
      "step": 3386,
      "task_loss": 0.14617042243480682
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7786095749296322,
      "compression_loss": 0.0,
      "distillation_loss": 0.1770530492067337,
      "epoch": 3.22,
      "learning_rate": 2.8203722486090168e-05,
      "loss": 0.1791,
      "step": 3387,
      "task_loss": 0.19720126688480377
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7786223098519434,
      "compression_loss": 0.0,
      "distillation_loss": 0.27764710783958435,
      "epoch": 3.22,
      "learning_rate": 2.8193154802833803e-05,
      "loss": 0.2611,
      "step": 3388,
      "task_loss": 0.11260776221752167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7786350397187077,
      "compression_loss": 0.0,
      "distillation_loss": 0.052394554018974304,
      "epoch": 3.22,
      "learning_rate": 2.818258653952549e-05,
      "loss": 0.0817,
      "step": 3389,
      "task_loss": 0.3452882766723633
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7786477645309291,
      "compression_loss": 0.0,
      "distillation_loss": 0.023975854739546776,
      "epoch": 3.22,
      "learning_rate": 2.8172017698085013e-05,
      "loss": 0.0298,
      "step": 3390,
      "task_loss": 0.08264005184173584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7786604842896112,
      "compression_loss": 0.0,
      "distillation_loss": 0.16987451910972595,
      "epoch": 3.22,
      "learning_rate": 2.816144828043224e-05,
      "loss": 0.1559,
      "step": 3391,
      "task_loss": 0.030469371005892754
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7786731989957576,
      "compression_loss": 0.0,
      "distillation_loss": 0.08702673763036728,
      "epoch": 3.22,
      "learning_rate": 2.8150878288487155e-05,
      "loss": 0.0822,
      "step": 3392,
      "task_loss": 0.03916516155004501
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.778685908650372,
      "compression_loss": 0.0,
      "distillation_loss": 0.030004359781742096,
      "epoch": 3.22,
      "learning_rate": 2.8140307724169857e-05,
      "loss": 0.0351,
      "step": 3393,
      "task_loss": 0.08116083592176437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7786986132544582,
      "compression_loss": 0.0,
      "distillation_loss": 0.18179570138454437,
      "epoch": 3.22,
      "learning_rate": 2.812973658940054e-05,
      "loss": 0.199,
      "step": 3394,
      "task_loss": 0.3536381423473358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7787113128090197,
      "compression_loss": 0.0,
      "distillation_loss": 0.1652180403470993,
      "epoch": 3.22,
      "learning_rate": 2.8119164886099504e-05,
      "loss": 0.1694,
      "step": 3395,
      "task_loss": 0.2065887749195099
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7787240073150603,
      "compression_loss": 0.0,
      "distillation_loss": 0.10243000090122223,
      "epoch": 3.23,
      "learning_rate": 2.8108592616187133e-05,
      "loss": 0.1072,
      "step": 3396,
      "task_loss": 0.15036579966545105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7787366967735837,
      "compression_loss": 0.0,
      "distillation_loss": 0.12697230279445648,
      "epoch": 3.23,
      "learning_rate": 2.8098019781583944e-05,
      "loss": 0.1212,
      "step": 3397,
      "task_loss": 0.06895315647125244
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7787493811855934,
      "compression_loss": 0.0,
      "distillation_loss": 0.046805739402770996,
      "epoch": 3.23,
      "learning_rate": 2.8087446384210547e-05,
      "loss": 0.0473,
      "step": 3398,
      "task_loss": 0.05141756683588028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7787620605520934,
      "compression_loss": 0.0,
      "distillation_loss": 0.024082593619823456,
      "epoch": 3.23,
      "learning_rate": 2.8076872425987637e-05,
      "loss": 0.0225,
      "step": 3399,
      "task_loss": 0.00831957720220089
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7787747348740871,
      "compression_loss": 0.0,
      "distillation_loss": 0.14285123348236084,
      "epoch": 3.23,
      "learning_rate": 2.8066297908836043e-05,
      "loss": 0.1427,
      "step": 3400,
      "task_loss": 0.14182282984256744
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7787874041525783,
      "compression_loss": 0.0,
      "distillation_loss": 0.08242295682430267,
      "epoch": 3.23,
      "learning_rate": 2.8055722834676658e-05,
      "loss": 0.0958,
      "step": 3401,
      "task_loss": 0.21660152077674866
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7788000683885706,
      "compression_loss": 0.0,
      "distillation_loss": 0.038724031299352646,
      "epoch": 3.23,
      "learning_rate": 2.804514720543051e-05,
      "loss": 0.0355,
      "step": 3402,
      "task_loss": 0.006068244576454163
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7788127275830679,
      "compression_loss": 0.0,
      "distillation_loss": 0.08571316301822662,
      "epoch": 3.23,
      "learning_rate": 2.80345710230187e-05,
      "loss": 0.0914,
      "step": 3403,
      "task_loss": 0.14219190180301666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7788253817370736,
      "compression_loss": 0.0,
      "distillation_loss": 0.07485407590866089,
      "epoch": 3.23,
      "learning_rate": 2.802399428936246e-05,
      "loss": 0.0825,
      "step": 3404,
      "task_loss": 0.15130427479743958
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7788380308515915,
      "compression_loss": 0.0,
      "distillation_loss": 0.09159151464700699,
      "epoch": 3.23,
      "learning_rate": 2.8013417006383076e-05,
      "loss": 0.09,
      "step": 3405,
      "task_loss": 0.07594650983810425
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7788506749276254,
      "compression_loss": 0.0,
      "distillation_loss": 0.0715874582529068,
      "epoch": 3.23,
      "learning_rate": 2.8002839176001987e-05,
      "loss": 0.0843,
      "step": 3406,
      "task_loss": 0.19857263565063477
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7788633139661788,
      "compression_loss": 0.0,
      "distillation_loss": 0.13416993618011475,
      "epoch": 3.24,
      "learning_rate": 2.799226080014071e-05,
      "loss": 0.1421,
      "step": 3407,
      "task_loss": 0.21369251608848572
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7788759479682554,
      "compression_loss": 0.0,
      "distillation_loss": 0.09951282292604446,
      "epoch": 3.24,
      "learning_rate": 2.7981681880720838e-05,
      "loss": 0.0955,
      "step": 3408,
      "task_loss": 0.058943700045347214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.778888576934859,
      "compression_loss": 0.0,
      "distillation_loss": 0.3170393109321594,
      "epoch": 3.24,
      "learning_rate": 2.7971102419664103e-05,
      "loss": 0.3122,
      "step": 3409,
      "task_loss": 0.2684779763221741
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7789012008669933,
      "compression_loss": 0.0,
      "distillation_loss": 0.04012390226125717,
      "epoch": 3.24,
      "learning_rate": 2.7960522418892288e-05,
      "loss": 0.0552,
      "step": 3410,
      "task_loss": 0.1905803382396698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7789138197656618,
      "compression_loss": 0.0,
      "distillation_loss": 0.08090061694383621,
      "epoch": 3.24,
      "learning_rate": 2.794994188032733e-05,
      "loss": 0.0841,
      "step": 3411,
      "task_loss": 0.11296429485082626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7789264336318683,
      "compression_loss": 0.0,
      "distillation_loss": 0.08891788125038147,
      "epoch": 3.24,
      "learning_rate": 2.7939360805891218e-05,
      "loss": 0.0852,
      "step": 3412,
      "task_loss": 0.05205078423023224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7789390424666165,
      "compression_loss": 0.0,
      "distillation_loss": 0.13765235245227814,
      "epoch": 3.24,
      "learning_rate": 2.7928779197506056e-05,
      "loss": 0.1313,
      "step": 3413,
      "task_loss": 0.07377782464027405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7789516462709101,
      "compression_loss": 0.0,
      "distillation_loss": 0.0779227688908577,
      "epoch": 3.24,
      "learning_rate": 2.7918197057094054e-05,
      "loss": 0.0794,
      "step": 3414,
      "task_loss": 0.09290515631437302
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7789642450457527,
      "compression_loss": 0.0,
      "distillation_loss": 0.15904906392097473,
      "epoch": 3.24,
      "learning_rate": 2.7907614386577497e-05,
      "loss": 0.1513,
      "step": 3415,
      "task_loss": 0.08202332258224487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.778976838792148,
      "compression_loss": 0.0,
      "distillation_loss": 0.05591778829693794,
      "epoch": 3.24,
      "learning_rate": 2.789703118787879e-05,
      "loss": 0.066,
      "step": 3416,
      "task_loss": 0.15640464425086975
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7789894275110996,
      "compression_loss": 0.0,
      "distillation_loss": 0.2858087122440338,
      "epoch": 3.25,
      "learning_rate": 2.7886447462920412e-05,
      "loss": 0.2745,
      "step": 3417,
      "task_loss": 0.17287422716617584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7790020112036113,
      "compression_loss": 0.0,
      "distillation_loss": 0.05627952516078949,
      "epoch": 3.25,
      "learning_rate": 2.787586321362495e-05,
      "loss": 0.0517,
      "step": 3418,
      "task_loss": 0.010653091594576836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7790145898706868,
      "compression_loss": 0.0,
      "distillation_loss": 0.15222644805908203,
      "epoch": 3.25,
      "learning_rate": 2.7865278441915082e-05,
      "loss": 0.1488,
      "step": 3419,
      "task_loss": 0.11802099645137787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7790271635133298,
      "compression_loss": 0.0,
      "distillation_loss": 0.05383795499801636,
      "epoch": 3.25,
      "learning_rate": 2.785469314971359e-05,
      "loss": 0.0516,
      "step": 3420,
      "task_loss": 0.03138484060764313
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.779039732132544,
      "compression_loss": 0.0,
      "distillation_loss": 0.03762197121977806,
      "epoch": 3.25,
      "learning_rate": 2.7844107338943343e-05,
      "loss": 0.047,
      "step": 3421,
      "task_loss": 0.13129746913909912
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7790522957293328,
      "compression_loss": 0.0,
      "distillation_loss": 0.13689181208610535,
      "epoch": 3.25,
      "learning_rate": 2.7833521011527293e-05,
      "loss": 0.1453,
      "step": 3422,
      "task_loss": 0.221421480178833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7790648543047002,
      "compression_loss": 0.0,
      "distillation_loss": 0.2846853733062744,
      "epoch": 3.25,
      "learning_rate": 2.782293416938851e-05,
      "loss": 0.2751,
      "step": 3423,
      "task_loss": 0.18841396272182465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7790774078596497,
      "compression_loss": 0.0,
      "distillation_loss": 0.05968084558844566,
      "epoch": 3.25,
      "learning_rate": 2.7812346814450135e-05,
      "loss": 0.078,
      "step": 3424,
      "task_loss": 0.2424488067626953
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7790899563951851,
      "compression_loss": 0.0,
      "distillation_loss": 0.08280298113822937,
      "epoch": 3.25,
      "learning_rate": 2.7801758948635414e-05,
      "loss": 0.0881,
      "step": 3425,
      "task_loss": 0.13536496460437775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7791024999123101,
      "compression_loss": 0.0,
      "distillation_loss": 0.11989252269268036,
      "epoch": 3.25,
      "learning_rate": 2.77911705738677e-05,
      "loss": 0.1164,
      "step": 3426,
      "task_loss": 0.08493325114250183
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7791150384120282,
      "compression_loss": 0.0,
      "distillation_loss": 0.1451410949230194,
      "epoch": 3.25,
      "learning_rate": 2.7780581692070395e-05,
      "loss": 0.1492,
      "step": 3427,
      "task_loss": 0.18578004837036133
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7791275718953432,
      "compression_loss": 0.0,
      "distillation_loss": 0.16686482727527618,
      "epoch": 3.26,
      "learning_rate": 2.7769992305167043e-05,
      "loss": 0.1621,
      "step": 3428,
      "task_loss": 0.11886778473854065
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7791401003632589,
      "compression_loss": 0.0,
      "distillation_loss": 0.057650670409202576,
      "epoch": 3.26,
      "learning_rate": 2.775940241508124e-05,
      "loss": 0.057,
      "step": 3429,
      "task_loss": 0.05127601698040962
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7791526238167787,
      "compression_loss": 0.0,
      "distillation_loss": 0.053712643682956696,
      "epoch": 3.26,
      "learning_rate": 2.774881202373671e-05,
      "loss": 0.064,
      "step": 3430,
      "task_loss": 0.1569092869758606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7791651422569066,
      "compression_loss": 0.0,
      "distillation_loss": 0.04081248864531517,
      "epoch": 3.26,
      "learning_rate": 2.773822113305723e-05,
      "loss": 0.0428,
      "step": 3431,
      "task_loss": 0.061016641557216644
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7791776556846461,
      "compression_loss": 0.0,
      "distillation_loss": 0.06572188436985016,
      "epoch": 3.26,
      "learning_rate": 2.7727629744966695e-05,
      "loss": 0.067,
      "step": 3432,
      "task_loss": 0.07893720269203186
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7791901641010008,
      "compression_loss": 0.0,
      "distillation_loss": 0.02726941555738449,
      "epoch": 3.26,
      "learning_rate": 2.7717037861389082e-05,
      "loss": 0.0256,
      "step": 3433,
      "task_loss": 0.010105656459927559
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7792026675069746,
      "compression_loss": 0.0,
      "distillation_loss": 0.030473362654447556,
      "epoch": 3.26,
      "learning_rate": 2.7706445484248454e-05,
      "loss": 0.0291,
      "step": 3434,
      "task_loss": 0.017233194783329964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7792151659035711,
      "compression_loss": 0.0,
      "distillation_loss": 0.2087365984916687,
      "epoch": 3.26,
      "learning_rate": 2.769585261546897e-05,
      "loss": 0.2018,
      "step": 3435,
      "task_loss": 0.13944987952709198
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7792276592917939,
      "compression_loss": 0.0,
      "distillation_loss": 0.04427679255604744,
      "epoch": 3.26,
      "learning_rate": 2.768525925697487e-05,
      "loss": 0.0431,
      "step": 3436,
      "task_loss": 0.03255880996584892
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7792401476726467,
      "compression_loss": 0.0,
      "distillation_loss": 0.10959358513355255,
      "epoch": 3.26,
      "learning_rate": 2.76746654106905e-05,
      "loss": 0.1043,
      "step": 3437,
      "task_loss": 0.05700678750872612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7792526310471333,
      "compression_loss": 0.0,
      "distillation_loss": 0.18829983472824097,
      "epoch": 3.26,
      "learning_rate": 2.7664071078540282e-05,
      "loss": 0.1821,
      "step": 3438,
      "task_loss": 0.12653696537017822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7792651094162573,
      "compression_loss": 0.0,
      "distillation_loss": 0.12829965353012085,
      "epoch": 3.27,
      "learning_rate": 2.7653476262448713e-05,
      "loss": 0.1339,
      "step": 3439,
      "task_loss": 0.18381281197071075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7792775827810223,
      "compression_loss": 0.0,
      "distillation_loss": 0.06626725196838379,
      "epoch": 3.27,
      "learning_rate": 2.76428809643404e-05,
      "loss": 0.062,
      "step": 3440,
      "task_loss": 0.023818595334887505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7792900511424321,
      "compression_loss": 0.0,
      "distillation_loss": 0.03690072149038315,
      "epoch": 3.27,
      "learning_rate": 2.763228518614004e-05,
      "loss": 0.0498,
      "step": 3441,
      "task_loss": 0.16629740595817566
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7793025145014904,
      "compression_loss": 0.0,
      "distillation_loss": 0.04804535210132599,
      "epoch": 3.27,
      "learning_rate": 2.7621688929772393e-05,
      "loss": 0.0533,
      "step": 3442,
      "task_loss": 0.10066729038953781
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7793149728592008,
      "compression_loss": 0.0,
      "distillation_loss": 0.1166885569691658,
      "epoch": 3.27,
      "learning_rate": 2.761109219716233e-05,
      "loss": 0.1143,
      "step": 3443,
      "task_loss": 0.09277547895908356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7793274262165669,
      "compression_loss": 0.0,
      "distillation_loss": 0.10858124494552612,
      "epoch": 3.27,
      "learning_rate": 2.760049499023479e-05,
      "loss": 0.113,
      "step": 3444,
      "task_loss": 0.15251390635967255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7793398745745926,
      "compression_loss": 0.0,
      "distillation_loss": 0.03859622776508331,
      "epoch": 3.27,
      "learning_rate": 2.7589897310914814e-05,
      "loss": 0.0443,
      "step": 3445,
      "task_loss": 0.09562104940414429
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7793523179342815,
      "compression_loss": 0.0,
      "distillation_loss": 0.030183183029294014,
      "epoch": 3.27,
      "learning_rate": 2.7579299161127513e-05,
      "loss": 0.0287,
      "step": 3446,
      "task_loss": 0.01539078913629055
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7793647562966373,
      "compression_loss": 0.0,
      "distillation_loss": 0.06578610837459564,
      "epoch": 3.27,
      "learning_rate": 2.756870054279811e-05,
      "loss": 0.0839,
      "step": 3447,
      "task_loss": 0.24706564843654633
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7793771896626636,
      "compression_loss": 0.0,
      "distillation_loss": 0.2527296543121338,
      "epoch": 3.27,
      "learning_rate": 2.755810145785187e-05,
      "loss": 0.2617,
      "step": 3448,
      "task_loss": 0.3425305485725403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.779389618033364,
      "compression_loss": 0.0,
      "distillation_loss": 0.10953805595636368,
      "epoch": 3.28,
      "learning_rate": 2.754750190821418e-05,
      "loss": 0.1019,
      "step": 3449,
      "task_loss": 0.03293545916676521
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7794020414097425,
      "compression_loss": 0.0,
      "distillation_loss": 0.2223641574382782,
      "epoch": 3.28,
      "learning_rate": 2.753690189581051e-05,
      "loss": 0.2129,
      "step": 3450,
      "task_loss": 0.12748362123966217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7794144597928024,
      "compression_loss": 0.0,
      "distillation_loss": 0.07554468512535095,
      "epoch": 3.28,
      "learning_rate": 2.752630142256638e-05,
      "loss": 0.0726,
      "step": 3451,
      "task_loss": 0.045700106769800186
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7794268731835476,
      "compression_loss": 0.0,
      "distillation_loss": 0.0460401251912117,
      "epoch": 3.28,
      "learning_rate": 2.7515700490407443e-05,
      "loss": 0.0571,
      "step": 3452,
      "task_loss": 0.15684375166893005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7794392815829819,
      "compression_loss": 0.0,
      "distillation_loss": 0.07692290842533112,
      "epoch": 3.28,
      "learning_rate": 2.7505099101259386e-05,
      "loss": 0.0725,
      "step": 3453,
      "task_loss": 0.03274589031934738
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7794516849921087,
      "compression_loss": 0.0,
      "distillation_loss": 0.1473102867603302,
      "epoch": 3.28,
      "learning_rate": 2.749449725704802e-05,
      "loss": 0.1434,
      "step": 3454,
      "task_loss": 0.1077522411942482
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7794640834119319,
      "compression_loss": 0.0,
      "distillation_loss": 0.10021187365055084,
      "epoch": 3.28,
      "learning_rate": 2.748389495969921e-05,
      "loss": 0.1095,
      "step": 3455,
      "task_loss": 0.19332432746887207
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.779476476843455,
      "compression_loss": 0.0,
      "distillation_loss": 0.03207477182149887,
      "epoch": 3.28,
      "learning_rate": 2.747329221113891e-05,
      "loss": 0.0303,
      "step": 3456,
      "task_loss": 0.01403326727449894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7794888652876819,
      "compression_loss": 0.0,
      "distillation_loss": 0.1490982323884964,
      "epoch": 3.28,
      "learning_rate": 2.7462689013293176e-05,
      "loss": 0.1534,
      "step": 3457,
      "task_loss": 0.1916988492012024
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7795012487456161,
      "compression_loss": 0.0,
      "distillation_loss": 0.07416212558746338,
      "epoch": 3.28,
      "learning_rate": 2.745208536808812e-05,
      "loss": 0.0814,
      "step": 3458,
      "task_loss": 0.14629532396793365
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7795136272182613,
      "compression_loss": 0.0,
      "distillation_loss": 0.3036563992500305,
      "epoch": 3.28,
      "learning_rate": 2.7441481277449954e-05,
      "loss": 0.2975,
      "step": 3459,
      "task_loss": 0.24229754507541656
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7795260007066213,
      "compression_loss": 0.0,
      "distillation_loss": 0.05779052898287773,
      "epoch": 3.29,
      "learning_rate": 2.743087674330495e-05,
      "loss": 0.0536,
      "step": 3460,
      "task_loss": 0.016016999259591103
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7795383692116996,
      "compression_loss": 0.0,
      "distillation_loss": 0.05794607847929001,
      "epoch": 3.29,
      "learning_rate": 2.742027176757948e-05,
      "loss": 0.054,
      "step": 3461,
      "task_loss": 0.018953843042254448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7795507327345002,
      "compression_loss": 0.0,
      "distillation_loss": 0.13176500797271729,
      "epoch": 3.29,
      "learning_rate": 2.7409666352199986e-05,
      "loss": 0.1273,
      "step": 3462,
      "task_loss": 0.08700872212648392
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7795630912760264,
      "compression_loss": 0.0,
      "distillation_loss": 0.03117290511727333,
      "epoch": 3.29,
      "learning_rate": 2.7399060499092992e-05,
      "loss": 0.0383,
      "step": 3463,
      "task_loss": 0.10199519991874695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.779575444837282,
      "compression_loss": 0.0,
      "distillation_loss": 0.03322270140051842,
      "epoch": 3.29,
      "learning_rate": 2.7388454210185115e-05,
      "loss": 0.0369,
      "step": 3464,
      "task_loss": 0.0694967657327652
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7795877934192709,
      "compression_loss": 0.0,
      "distillation_loss": 0.06944756209850311,
      "epoch": 3.29,
      "learning_rate": 2.7377847487403018e-05,
      "loss": 0.0634,
      "step": 3465,
      "task_loss": 0.008718544617295265
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7796001370229966,
      "compression_loss": 0.0,
      "distillation_loss": 0.07945192605257034,
      "epoch": 3.29,
      "learning_rate": 2.736724033267347e-05,
      "loss": 0.0837,
      "step": 3466,
      "task_loss": 0.12163371592760086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7796124756494627,
      "compression_loss": 0.0,
      "distillation_loss": 0.04697978496551514,
      "epoch": 3.29,
      "learning_rate": 2.7356632747923322e-05,
      "loss": 0.0436,
      "step": 3467,
      "task_loss": 0.013647403568029404
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.779624809299673,
      "compression_loss": 0.0,
      "distillation_loss": 0.10399874299764633,
      "epoch": 3.29,
      "learning_rate": 2.7346024735079486e-05,
      "loss": 0.1006,
      "step": 3468,
      "task_loss": 0.06967251002788544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7796371379746312,
      "compression_loss": 0.0,
      "distillation_loss": 0.20231911540031433,
      "epoch": 3.29,
      "learning_rate": 2.7335416296068962e-05,
      "loss": 0.204,
      "step": 3469,
      "task_loss": 0.21951937675476074
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.779649461675341,
      "compression_loss": 0.0,
      "distillation_loss": 0.09952305257320404,
      "epoch": 3.3,
      "learning_rate": 2.7324807432818805e-05,
      "loss": 0.093,
      "step": 3470,
      "task_loss": 0.033969540148973465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7796617804028059,
      "compression_loss": 0.0,
      "distillation_loss": 0.2023647427558899,
      "epoch": 3.3,
      "learning_rate": 2.731419814725619e-05,
      "loss": 0.1988,
      "step": 3471,
      "task_loss": 0.16686657071113586
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7796740941580298,
      "compression_loss": 0.0,
      "distillation_loss": 0.019200827926397324,
      "epoch": 3.3,
      "learning_rate": 2.730358844130834e-05,
      "loss": 0.0176,
      "step": 3472,
      "task_loss": 0.0035965926945209503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7796864029420163,
      "compression_loss": 0.0,
      "distillation_loss": 0.10850808024406433,
      "epoch": 3.3,
      "learning_rate": 2.729297831690255e-05,
      "loss": 0.1036,
      "step": 3473,
      "task_loss": 0.059668056666851044
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.779698706755769,
      "compression_loss": 0.0,
      "distillation_loss": 0.1171179711818695,
      "epoch": 3.3,
      "learning_rate": 2.728236777596621e-05,
      "loss": 0.1114,
      "step": 3474,
      "task_loss": 0.0599634051322937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7797110056002917,
      "compression_loss": 0.0,
      "distillation_loss": 0.023795567452907562,
      "epoch": 3.3,
      "learning_rate": 2.7271756820426763e-05,
      "loss": 0.0218,
      "step": 3475,
      "task_loss": 0.003354804590344429
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7797232994765881,
      "compression_loss": 0.0,
      "distillation_loss": 0.05289144068956375,
      "epoch": 3.3,
      "learning_rate": 2.7261145452211763e-05,
      "loss": 0.0625,
      "step": 3476,
      "task_loss": 0.14865252375602722
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7797355883856617,
      "compression_loss": 0.0,
      "distillation_loss": 0.26715028285980225,
      "epoch": 3.3,
      "learning_rate": 2.725053367324879e-05,
      "loss": 0.271,
      "step": 3477,
      "task_loss": 0.3059850335121155
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7797478723285164,
      "compression_loss": 0.0,
      "distillation_loss": 0.03925115987658501,
      "epoch": 3.3,
      "learning_rate": 2.723992148546554e-05,
      "loss": 0.0448,
      "step": 3478,
      "task_loss": 0.0945422425866127
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7797601513061558,
      "compression_loss": 0.0,
      "distillation_loss": 0.10821853578090668,
      "epoch": 3.3,
      "learning_rate": 2.7229308890789767e-05,
      "loss": 0.1106,
      "step": 3479,
      "task_loss": 0.13163337111473083
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7797724253195835,
      "compression_loss": 0.0,
      "distillation_loss": 0.026758499443531036,
      "epoch": 3.3,
      "learning_rate": 2.7218695891149293e-05,
      "loss": 0.0359,
      "step": 3480,
      "task_loss": 0.11779153347015381
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7797846943698032,
      "compression_loss": 0.0,
      "distillation_loss": 0.04860454425215721,
      "epoch": 3.31,
      "learning_rate": 2.720808248847203e-05,
      "loss": 0.0449,
      "step": 3481,
      "task_loss": 0.011549010872840881
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7797969584578188,
      "compression_loss": 0.0,
      "distillation_loss": 0.11690667271614075,
      "epoch": 3.31,
      "learning_rate": 2.719746868468595e-05,
      "loss": 0.1101,
      "step": 3482,
      "task_loss": 0.04849759489297867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7798092175846337,
      "compression_loss": 0.0,
      "distillation_loss": 0.13424092531204224,
      "epoch": 3.31,
      "learning_rate": 2.7186854481719092e-05,
      "loss": 0.1333,
      "step": 3483,
      "task_loss": 0.12473595142364502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7798214717512517,
      "compression_loss": 0.0,
      "distillation_loss": 0.07486072927713394,
      "epoch": 3.31,
      "learning_rate": 2.7176239881499595e-05,
      "loss": 0.0961,
      "step": 3484,
      "task_loss": 0.28720682859420776
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7798337209586765,
      "compression_loss": 0.0,
      "distillation_loss": 0.1800866276025772,
      "epoch": 3.31,
      "learning_rate": 2.716562488595563e-05,
      "loss": 0.1734,
      "step": 3485,
      "task_loss": 0.11275843530893326
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7798459652079118,
      "compression_loss": 0.0,
      "distillation_loss": 0.045273929834365845,
      "epoch": 3.31,
      "learning_rate": 2.715500949701549e-05,
      "loss": 0.0426,
      "step": 3486,
      "task_loss": 0.0184002872556448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7798582044999612,
      "compression_loss": 0.0,
      "distillation_loss": 0.07528147101402283,
      "epoch": 3.31,
      "learning_rate": 2.7144393716607486e-05,
      "loss": 0.0723,
      "step": 3487,
      "task_loss": 0.04534187912940979
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7798704388358284,
      "compression_loss": 0.0,
      "distillation_loss": 0.1249753087759018,
      "epoch": 3.31,
      "learning_rate": 2.713377754666004e-05,
      "loss": 0.1209,
      "step": 3488,
      "task_loss": 0.08464540541172028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7798826682165172,
      "compression_loss": 0.0,
      "distillation_loss": 0.2507629990577698,
      "epoch": 3.31,
      "learning_rate": 2.712316098910162e-05,
      "loss": 0.2497,
      "step": 3489,
      "task_loss": 0.23983043432235718
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.779894892643031,
      "compression_loss": 0.0,
      "distillation_loss": 0.017407327890396118,
      "epoch": 3.31,
      "learning_rate": 2.711254404586079e-05,
      "loss": 0.0162,
      "step": 3490,
      "task_loss": 0.004980321973562241
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7799071121163739,
      "compression_loss": 0.0,
      "distillation_loss": 0.16080042719841003,
      "epoch": 3.32,
      "learning_rate": 2.7101926718866156e-05,
      "loss": 0.1569,
      "step": 3491,
      "task_loss": 0.12228196114301682
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7799193266375491,
      "compression_loss": 0.0,
      "distillation_loss": 0.20948654413223267,
      "epoch": 3.32,
      "learning_rate": 2.7091309010046408e-05,
      "loss": 0.2093,
      "step": 3492,
      "task_loss": 0.2079431563615799
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7799315362075607,
      "compression_loss": 0.0,
      "distillation_loss": 0.11191666126251221,
      "epoch": 3.32,
      "learning_rate": 2.708069092133031e-05,
      "loss": 0.1132,
      "step": 3493,
      "task_loss": 0.12430918961763382
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7799437408274122,
      "compression_loss": 0.0,
      "distillation_loss": 0.11181183904409409,
      "epoch": 3.32,
      "learning_rate": 2.7070072454646683e-05,
      "loss": 0.106,
      "step": 3494,
      "task_loss": 0.05379234626889229
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7799559404981073,
      "compression_loss": 0.0,
      "distillation_loss": 0.09605347365140915,
      "epoch": 3.32,
      "learning_rate": 2.7059453611924433e-05,
      "loss": 0.0923,
      "step": 3495,
      "task_loss": 0.058432966470718384
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7799681352206498,
      "compression_loss": 0.0,
      "distillation_loss": 0.06996072828769684,
      "epoch": 3.32,
      "learning_rate": 2.7048834395092505e-05,
      "loss": 0.0752,
      "step": 3496,
      "task_loss": 0.12205319851636887
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7799803249960431,
      "compression_loss": 0.0,
      "distillation_loss": 0.09135409444570541,
      "epoch": 3.32,
      "learning_rate": 2.7038214806079948e-05,
      "loss": 0.0947,
      "step": 3497,
      "task_loss": 0.12439997494220734
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7799925098252911,
      "compression_loss": 0.0,
      "distillation_loss": 0.04232946038246155,
      "epoch": 3.32,
      "learning_rate": 2.702759484681585e-05,
      "loss": 0.0455,
      "step": 3498,
      "task_loss": 0.07386565208435059
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7800046897093974,
      "compression_loss": 0.0,
      "distillation_loss": 0.06012481451034546,
      "epoch": 3.32,
      "learning_rate": 2.701697451922939e-05,
      "loss": 0.0557,
      "step": 3499,
      "task_loss": 0.015828527510166168
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7800168646493657,
      "compression_loss": 0.0,
      "distillation_loss": 0.13709309697151184,
      "epoch": 3.32,
      "learning_rate": 2.7006353825249792e-05,
      "loss": 0.1315,
      "step": 3500,
      "task_loss": 0.08080706745386124
    },
    {
      "epoch": 3.32,
      "eval_accuracy": 0.8967889908256881,
      "eval_loss": 0.41444137692451477,
      "eval_runtime": 18.5229,
      "eval_samples_per_second": 47.077,
      "eval_steps_per_second": 5.885,
      "step": 3500
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7800290346461998,
      "compression_loss": 0.0,
      "distillation_loss": 0.05133210867643356,
      "epoch": 3.32,
      "learning_rate": 2.6995732766806354e-05,
      "loss": 0.0478,
      "step": 3501,
      "task_loss": 0.015585673972964287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7800411997009031,
      "compression_loss": 0.0,
      "distillation_loss": 0.04098585247993469,
      "epoch": 3.33,
      "learning_rate": 2.6985111345828452e-05,
      "loss": 0.0458,
      "step": 3502,
      "task_loss": 0.0887080505490303
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7800533598144795,
      "compression_loss": 0.0,
      "distillation_loss": 0.07880129665136337,
      "epoch": 3.33,
      "learning_rate": 2.6974489564245513e-05,
      "loss": 0.0739,
      "step": 3503,
      "task_loss": 0.030195550993084908
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7800655149879328,
      "compression_loss": 0.0,
      "distillation_loss": 0.07193966954946518,
      "epoch": 3.33,
      "learning_rate": 2.6963867423987032e-05,
      "loss": 0.0733,
      "step": 3504,
      "task_loss": 0.08535440266132355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7800776652222663,
      "compression_loss": 0.0,
      "distillation_loss": 0.23703370988368988,
      "epoch": 3.33,
      "learning_rate": 2.695324492698258e-05,
      "loss": 0.2278,
      "step": 3505,
      "task_loss": 0.14506091177463531
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.780089810518484,
      "compression_loss": 0.0,
      "distillation_loss": 0.27318769693374634,
      "epoch": 3.33,
      "learning_rate": 2.694262207516178e-05,
      "loss": 0.2656,
      "step": 3506,
      "task_loss": 0.1974327713251114
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7801019508775895,
      "compression_loss": 0.0,
      "distillation_loss": 0.03283765912055969,
      "epoch": 3.33,
      "learning_rate": 2.6931998870454327e-05,
      "loss": 0.0311,
      "step": 3507,
      "task_loss": 0.015567878261208534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7801140863005865,
      "compression_loss": 0.0,
      "distillation_loss": 0.08824481815099716,
      "epoch": 3.33,
      "learning_rate": 2.692137531478997e-05,
      "loss": 0.0837,
      "step": 3508,
      "task_loss": 0.042351722717285156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7801262167884786,
      "compression_loss": 0.0,
      "distillation_loss": 0.10943618416786194,
      "epoch": 3.33,
      "learning_rate": 2.6910751410098532e-05,
      "loss": 0.1129,
      "step": 3509,
      "task_loss": 0.14431238174438477
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7801383423422695,
      "compression_loss": 0.0,
      "distillation_loss": 0.10514090955257416,
      "epoch": 3.33,
      "learning_rate": 2.6900127158309903e-05,
      "loss": 0.1055,
      "step": 3510,
      "task_loss": 0.10892733186483383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.780150462962963,
      "compression_loss": 0.0,
      "distillation_loss": 0.1573885828256607,
      "epoch": 3.33,
      "learning_rate": 2.688950256135402e-05,
      "loss": 0.1667,
      "step": 3511,
      "task_loss": 0.2509816586971283
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7801625786515627,
      "compression_loss": 0.0,
      "distillation_loss": 0.0467422753572464,
      "epoch": 3.34,
      "learning_rate": 2.6878877621160904e-05,
      "loss": 0.0526,
      "step": 3512,
      "task_loss": 0.10526955127716064
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7801746894090722,
      "compression_loss": 0.0,
      "distillation_loss": 0.0356249138712883,
      "epoch": 3.34,
      "learning_rate": 2.686825233966061e-05,
      "loss": 0.0428,
      "step": 3513,
      "task_loss": 0.10746340453624725
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7801867952364953,
      "compression_loss": 0.0,
      "distillation_loss": 0.07927364856004715,
      "epoch": 3.34,
      "learning_rate": 2.6857626718783285e-05,
      "loss": 0.0928,
      "step": 3514,
      "task_loss": 0.21453584730625153
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7801988961348355,
      "compression_loss": 0.0,
      "distillation_loss": 0.21596963703632355,
      "epoch": 3.34,
      "learning_rate": 2.6847000760459118e-05,
      "loss": 0.2164,
      "step": 3515,
      "task_loss": 0.22040767967700958
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7802109921050968,
      "compression_loss": 0.0,
      "distillation_loss": 0.04428340867161751,
      "epoch": 3.34,
      "learning_rate": 2.683637446661837e-05,
      "loss": 0.0504,
      "step": 3516,
      "task_loss": 0.10571881383657455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7802230831482826,
      "compression_loss": 0.0,
      "distillation_loss": 0.020135464146733284,
      "epoch": 3.34,
      "learning_rate": 2.6825747839191362e-05,
      "loss": 0.0284,
      "step": 3517,
      "task_loss": 0.10235545039176941
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7802351692653968,
      "compression_loss": 0.0,
      "distillation_loss": 0.09116624295711517,
      "epoch": 3.34,
      "learning_rate": 2.681512088010845e-05,
      "loss": 0.0861,
      "step": 3518,
      "task_loss": 0.04042452201247215
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7802472504574429,
      "compression_loss": 0.0,
      "distillation_loss": 0.21908923983573914,
      "epoch": 3.34,
      "learning_rate": 2.6804493591300105e-05,
      "loss": 0.2111,
      "step": 3519,
      "task_loss": 0.13956592977046967
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7802593267254246,
      "compression_loss": 0.0,
      "distillation_loss": 0.09275214374065399,
      "epoch": 3.34,
      "learning_rate": 2.6793865974696803e-05,
      "loss": 0.0882,
      "step": 3520,
      "task_loss": 0.046760689467191696
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7802713980703457,
      "compression_loss": 0.0,
      "distillation_loss": 0.031631097197532654,
      "epoch": 3.34,
      "learning_rate": 2.67832380322291e-05,
      "loss": 0.0295,
      "step": 3521,
      "task_loss": 0.01028413511812687
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7802834644932098,
      "compression_loss": 0.0,
      "distillation_loss": 0.09307868778705597,
      "epoch": 3.34,
      "learning_rate": 2.6772609765827627e-05,
      "loss": 0.1006,
      "step": 3522,
      "task_loss": 0.16859915852546692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7802955259950206,
      "compression_loss": 0.0,
      "distillation_loss": 0.07208751142024994,
      "epoch": 3.35,
      "learning_rate": 2.6761981177423052e-05,
      "loss": 0.0748,
      "step": 3523,
      "task_loss": 0.09921170026063919
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7803075825767818,
      "compression_loss": 0.0,
      "distillation_loss": 0.05329529941082001,
      "epoch": 3.35,
      "learning_rate": 2.6751352268946118e-05,
      "loss": 0.0492,
      "step": 3524,
      "task_loss": 0.012433096766471863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7803196342394971,
      "compression_loss": 0.0,
      "distillation_loss": 0.13451752066612244,
      "epoch": 3.35,
      "learning_rate": 2.6740723042327598e-05,
      "loss": 0.1327,
      "step": 3525,
      "task_loss": 0.11605449765920639
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.78033168098417,
      "compression_loss": 0.0,
      "distillation_loss": 0.08334813266992569,
      "epoch": 3.35,
      "learning_rate": 2.673009349949836e-05,
      "loss": 0.1004,
      "step": 3526,
      "task_loss": 0.2539750039577484
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7803437228118045,
      "compression_loss": 0.0,
      "distillation_loss": 0.11693807691335678,
      "epoch": 3.35,
      "learning_rate": 2.6719463642389302e-05,
      "loss": 0.1105,
      "step": 3527,
      "task_loss": 0.05235512554645538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.780355759723404,
      "compression_loss": 0.0,
      "distillation_loss": 0.04535358399152756,
      "epoch": 3.35,
      "learning_rate": 2.6708833472931394e-05,
      "loss": 0.0491,
      "step": 3528,
      "task_loss": 0.08303047716617584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7803677917199723,
      "compression_loss": 0.0,
      "distillation_loss": 0.058416105806827545,
      "epoch": 3.35,
      "learning_rate": 2.669820299305566e-05,
      "loss": 0.0618,
      "step": 3529,
      "task_loss": 0.09264393150806427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7803798188025132,
      "compression_loss": 0.0,
      "distillation_loss": 0.0770866721868515,
      "epoch": 3.35,
      "learning_rate": 2.6687572204693174e-05,
      "loss": 0.0906,
      "step": 3530,
      "task_loss": 0.2119910567998886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7803918409720301,
      "compression_loss": 0.0,
      "distillation_loss": 0.042178817093372345,
      "epoch": 3.35,
      "learning_rate": 2.667694110977506e-05,
      "loss": 0.0418,
      "step": 3531,
      "task_loss": 0.03879820555448532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7804038582295268,
      "compression_loss": 0.0,
      "distillation_loss": 0.11523336172103882,
      "epoch": 3.35,
      "learning_rate": 2.6666309710232522e-05,
      "loss": 0.1115,
      "step": 3532,
      "task_loss": 0.078341543674469
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7804158705760071,
      "compression_loss": 0.0,
      "distillation_loss": 0.1296311616897583,
      "epoch": 3.36,
      "learning_rate": 2.6655678007996804e-05,
      "loss": 0.1225,
      "step": 3533,
      "task_loss": 0.057877760380506516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7804278780124746,
      "compression_loss": 0.0,
      "distillation_loss": 0.04751583933830261,
      "epoch": 3.36,
      "learning_rate": 2.66450460049992e-05,
      "loss": 0.0501,
      "step": 3534,
      "task_loss": 0.07348724454641342
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7804398805399331,
      "compression_loss": 0.0,
      "distillation_loss": 0.032740626484155655,
      "epoch": 3.36,
      "learning_rate": 2.6634413703171058e-05,
      "loss": 0.0306,
      "step": 3535,
      "task_loss": 0.011288370937108994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.780451878159386,
      "compression_loss": 0.0,
      "distillation_loss": 0.02693828195333481,
      "epoch": 3.36,
      "learning_rate": 2.6623781104443806e-05,
      "loss": 0.0299,
      "step": 3536,
      "task_loss": 0.05642160773277283
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7804638708718372,
      "compression_loss": 0.0,
      "distillation_loss": 0.1141064316034317,
      "epoch": 3.36,
      "learning_rate": 2.6613148210748894e-05,
      "loss": 0.1129,
      "step": 3537,
      "task_loss": 0.10202029347419739
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7804758586782903,
      "compression_loss": 0.0,
      "distillation_loss": 0.05455036461353302,
      "epoch": 3.36,
      "learning_rate": 2.6602515024017842e-05,
      "loss": 0.0595,
      "step": 3538,
      "task_loss": 0.1041443794965744
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.780487841579749,
      "compression_loss": 0.0,
      "distillation_loss": 0.03301164135336876,
      "epoch": 3.36,
      "learning_rate": 2.6591881546182216e-05,
      "loss": 0.0302,
      "step": 3539,
      "task_loss": 0.005180429667234421
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.780499819577217,
      "compression_loss": 0.0,
      "distillation_loss": 0.06954570859670639,
      "epoch": 3.36,
      "learning_rate": 2.6581247779173635e-05,
      "loss": 0.069,
      "step": 3540,
      "task_loss": 0.06367438286542892
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.780511792671698,
      "compression_loss": 0.0,
      "distillation_loss": 0.02071313187479973,
      "epoch": 3.36,
      "learning_rate": 2.6570613724923788e-05,
      "loss": 0.019,
      "step": 3541,
      "task_loss": 0.003429897129535675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7805237608641957,
      "compression_loss": 0.0,
      "distillation_loss": 0.01836605742573738,
      "epoch": 3.36,
      "learning_rate": 2.655997938536439e-05,
      "loss": 0.017,
      "step": 3542,
      "task_loss": 0.004780923947691917
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7805357241557137,
      "compression_loss": 0.0,
      "distillation_loss": 0.031529366970062256,
      "epoch": 3.36,
      "learning_rate": 2.654934476242723e-05,
      "loss": 0.0292,
      "step": 3543,
      "task_loss": 0.008400822058320045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7805476825472557,
      "compression_loss": 0.0,
      "distillation_loss": 0.026692230254411697,
      "epoch": 3.37,
      "learning_rate": 2.653870985804412e-05,
      "loss": 0.0248,
      "step": 3544,
      "task_loss": 0.007382074370980263
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7805596360398255,
      "compression_loss": 0.0,
      "distillation_loss": 0.036252401769161224,
      "epoch": 3.37,
      "learning_rate": 2.6528074674146963e-05,
      "loss": 0.0462,
      "step": 3545,
      "task_loss": 0.13550357520580292
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7805715846344266,
      "compression_loss": 0.0,
      "distillation_loss": 0.031404148787260056,
      "epoch": 3.37,
      "learning_rate": 2.6517439212667677e-05,
      "loss": 0.0292,
      "step": 3546,
      "task_loss": 0.009732730686664581
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7805835283320628,
      "compression_loss": 0.0,
      "distillation_loss": 0.08225059509277344,
      "epoch": 3.37,
      "learning_rate": 2.6506803475538256e-05,
      "loss": 0.0909,
      "step": 3547,
      "task_loss": 0.16853067278862
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7805954671337377,
      "compression_loss": 0.0,
      "distillation_loss": 0.028768369928002357,
      "epoch": 3.37,
      "learning_rate": 2.649616746469072e-05,
      "loss": 0.0268,
      "step": 3548,
      "task_loss": 0.009422983974218369
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7806074010404551,
      "compression_loss": 0.0,
      "distillation_loss": 0.11904192715883255,
      "epoch": 3.37,
      "learning_rate": 2.648553118205716e-05,
      "loss": 0.1248,
      "step": 3549,
      "task_loss": 0.17703823745250702
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7806193300532185,
      "compression_loss": 0.0,
      "distillation_loss": 0.03019685298204422,
      "epoch": 3.37,
      "learning_rate": 2.6474894629569713e-05,
      "loss": 0.0293,
      "step": 3550,
      "task_loss": 0.02082175202667713
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7806312541730318,
      "compression_loss": 0.0,
      "distillation_loss": 0.08796247839927673,
      "epoch": 3.37,
      "learning_rate": 2.6464257809160548e-05,
      "loss": 0.0828,
      "step": 3551,
      "task_loss": 0.03604867681860924
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7806431734008986,
      "compression_loss": 0.0,
      "distillation_loss": 0.03714694827795029,
      "epoch": 3.37,
      "learning_rate": 2.6453620722761896e-05,
      "loss": 0.0501,
      "step": 3552,
      "task_loss": 0.16653135418891907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7806550877378224,
      "compression_loss": 0.0,
      "distillation_loss": 0.15399909019470215,
      "epoch": 3.37,
      "learning_rate": 2.6442983372306045e-05,
      "loss": 0.1543,
      "step": 3553,
      "task_loss": 0.15673311054706573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7806669971848073,
      "compression_loss": 0.0,
      "distillation_loss": 0.23555459082126617,
      "epoch": 3.38,
      "learning_rate": 2.643234575972531e-05,
      "loss": 0.2352,
      "step": 3554,
      "task_loss": 0.2320120632648468
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7806789017428565,
      "compression_loss": 0.0,
      "distillation_loss": 0.0971142053604126,
      "epoch": 3.38,
      "learning_rate": 2.642170788695208e-05,
      "loss": 0.0907,
      "step": 3555,
      "task_loss": 0.03279740735888481
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.780690801412974,
      "compression_loss": 0.0,
      "distillation_loss": 0.05912064015865326,
      "epoch": 3.38,
      "learning_rate": 2.6411069755918755e-05,
      "loss": 0.0546,
      "step": 3556,
      "task_loss": 0.013454478234052658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7807026961961634,
      "compression_loss": 0.0,
      "distillation_loss": 0.15938124060630798,
      "epoch": 3.38,
      "learning_rate": 2.6400431368557815e-05,
      "loss": 0.1581,
      "step": 3557,
      "task_loss": 0.14631018042564392
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7807145860934284,
      "compression_loss": 0.0,
      "distillation_loss": 0.08248989284038544,
      "epoch": 3.38,
      "learning_rate": 2.6389792726801778e-05,
      "loss": 0.0976,
      "step": 3558,
      "task_loss": 0.23338665068149567
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7807264711057725,
      "compression_loss": 0.0,
      "distillation_loss": 0.03553071618080139,
      "epoch": 3.38,
      "learning_rate": 2.6379153832583186e-05,
      "loss": 0.0324,
      "step": 3559,
      "task_loss": 0.004168994724750519
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7807383512341997,
      "compression_loss": 0.0,
      "distillation_loss": 0.018167581409215927,
      "epoch": 3.38,
      "learning_rate": 2.6368514687834672e-05,
      "loss": 0.0167,
      "step": 3560,
      "task_loss": 0.003941915929317474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7807502264797134,
      "compression_loss": 0.0,
      "distillation_loss": 0.09068135917186737,
      "epoch": 3.38,
      "learning_rate": 2.6357875294488865e-05,
      "loss": 0.0855,
      "step": 3561,
      "task_loss": 0.03913940489292145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7807620968433175,
      "compression_loss": 0.0,
      "distillation_loss": 0.06539643555879593,
      "epoch": 3.38,
      "learning_rate": 2.6347235654478482e-05,
      "loss": 0.0653,
      "step": 3562,
      "task_loss": 0.06428472697734833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7807739623260155,
      "compression_loss": 0.0,
      "distillation_loss": 0.09735129773616791,
      "epoch": 3.38,
      "learning_rate": 2.6336595769736245e-05,
      "loss": 0.0938,
      "step": 3563,
      "task_loss": 0.06144891679286957
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7807858229288113,
      "compression_loss": 0.0,
      "distillation_loss": 0.1435181051492691,
      "epoch": 3.38,
      "learning_rate": 2.6325955642194948e-05,
      "loss": 0.1363,
      "step": 3564,
      "task_loss": 0.07164686918258667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7807976786527082,
      "compression_loss": 0.0,
      "distillation_loss": 0.05075103044509888,
      "epoch": 3.39,
      "learning_rate": 2.6315315273787428e-05,
      "loss": 0.0474,
      "step": 3565,
      "task_loss": 0.01732024922966957
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7808095294987103,
      "compression_loss": 0.0,
      "distillation_loss": 0.2770991325378418,
      "epoch": 3.39,
      "learning_rate": 2.630467466644655e-05,
      "loss": 0.2756,
      "step": 3566,
      "task_loss": 0.2625621557235718
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7808213754678212,
      "compression_loss": 0.0,
      "distillation_loss": 0.13243553042411804,
      "epoch": 3.39,
      "learning_rate": 2.629403382210524e-05,
      "loss": 0.1435,
      "step": 3567,
      "task_loss": 0.24289308488368988
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7808332165610443,
      "compression_loss": 0.0,
      "distillation_loss": 0.04136691614985466,
      "epoch": 3.39,
      "learning_rate": 2.628339274269645e-05,
      "loss": 0.0517,
      "step": 3568,
      "task_loss": 0.14519576728343964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7808450527793835,
      "compression_loss": 0.0,
      "distillation_loss": 0.014836801216006279,
      "epoch": 3.39,
      "learning_rate": 2.6272751430153186e-05,
      "loss": 0.0174,
      "step": 3569,
      "task_loss": 0.04046123847365379
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7808568841238426,
      "compression_loss": 0.0,
      "distillation_loss": 0.11791656911373138,
      "epoch": 3.39,
      "learning_rate": 2.62621098864085e-05,
      "loss": 0.1122,
      "step": 3570,
      "task_loss": 0.06102790683507919
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.780868710595425,
      "compression_loss": 0.0,
      "distillation_loss": 0.026029329746961594,
      "epoch": 3.39,
      "learning_rate": 2.6251468113395465e-05,
      "loss": 0.0323,
      "step": 3571,
      "task_loss": 0.08885947614908218
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7808805321951345,
      "compression_loss": 0.0,
      "distillation_loss": 0.07591719180345535,
      "epoch": 3.39,
      "learning_rate": 2.6240826113047235e-05,
      "loss": 0.0735,
      "step": 3572,
      "task_loss": 0.05175752192735672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.780892348923975,
      "compression_loss": 0.0,
      "distillation_loss": 0.030038170516490936,
      "epoch": 3.39,
      "learning_rate": 2.6230183887296955e-05,
      "loss": 0.0315,
      "step": 3573,
      "task_loss": 0.044650256633758545
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7809041607829498,
      "compression_loss": 0.0,
      "distillation_loss": 0.0887594223022461,
      "epoch": 3.39,
      "learning_rate": 2.6219541438077855e-05,
      "loss": 0.0841,
      "step": 3574,
      "task_loss": 0.04259561002254486
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7809159677730628,
      "compression_loss": 0.0,
      "distillation_loss": 0.10874880850315094,
      "epoch": 3.4,
      "learning_rate": 2.620889876732317e-05,
      "loss": 0.1032,
      "step": 3575,
      "task_loss": 0.05339481681585312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7809277698953176,
      "compression_loss": 0.0,
      "distillation_loss": 0.03946895897388458,
      "epoch": 3.4,
      "learning_rate": 2.6198255876966204e-05,
      "loss": 0.0368,
      "step": 3576,
      "task_loss": 0.012670749798417091
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7809395671507181,
      "compression_loss": 0.0,
      "distillation_loss": 0.2581176459789276,
      "epoch": 3.4,
      "learning_rate": 2.6187612768940293e-05,
      "loss": 0.2457,
      "step": 3577,
      "task_loss": 0.13422563672065735
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7809513595402676,
      "compression_loss": 0.0,
      "distillation_loss": 0.0735367089509964,
      "epoch": 3.4,
      "learning_rate": 2.61769694451788e-05,
      "loss": 0.0734,
      "step": 3578,
      "task_loss": 0.07223978638648987
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7809631470649702,
      "compression_loss": 0.0,
      "distillation_loss": 0.07702729851007462,
      "epoch": 3.4,
      "learning_rate": 2.616632590761514e-05,
      "loss": 0.078,
      "step": 3579,
      "task_loss": 0.08705680817365646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7809749297258293,
      "compression_loss": 0.0,
      "distillation_loss": 0.037445612251758575,
      "epoch": 3.4,
      "learning_rate": 2.615568215818276e-05,
      "loss": 0.0375,
      "step": 3580,
      "task_loss": 0.03814993426203728
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7809867075238486,
      "compression_loss": 0.0,
      "distillation_loss": 0.04116063937544823,
      "epoch": 3.4,
      "learning_rate": 2.6145038198815152e-05,
      "loss": 0.038,
      "step": 3581,
      "task_loss": 0.00981508381664753
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7809984804600318,
      "compression_loss": 0.0,
      "distillation_loss": 0.09640266001224518,
      "epoch": 3.4,
      "learning_rate": 2.6134394031445843e-05,
      "loss": 0.1068,
      "step": 3582,
      "task_loss": 0.20053042471408844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7810102485353828,
      "compression_loss": 0.0,
      "distillation_loss": 0.14251397550106049,
      "epoch": 3.4,
      "learning_rate": 2.6123749658008383e-05,
      "loss": 0.1474,
      "step": 3583,
      "task_loss": 0.1909736692905426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.781022011750905,
      "compression_loss": 0.0,
      "distillation_loss": 0.027869436889886856,
      "epoch": 3.4,
      "learning_rate": 2.6113105080436396e-05,
      "loss": 0.0299,
      "step": 3584,
      "task_loss": 0.04815257340669632
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7810337701076022,
      "compression_loss": 0.0,
      "distillation_loss": 0.048661135137081146,
      "epoch": 3.4,
      "learning_rate": 2.6102460300663506e-05,
      "loss": 0.0529,
      "step": 3585,
      "task_loss": 0.09094928950071335
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.781045523606478,
      "compression_loss": 0.0,
      "distillation_loss": 0.04565204679965973,
      "epoch": 3.41,
      "learning_rate": 2.60918153206234e-05,
      "loss": 0.0424,
      "step": 3586,
      "task_loss": 0.013126576319336891
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7810572722485363,
      "compression_loss": 0.0,
      "distillation_loss": 0.10344560444355011,
      "epoch": 3.41,
      "learning_rate": 2.6081170142249773e-05,
      "loss": 0.1083,
      "step": 3587,
      "task_loss": 0.1516730636358261
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7810690160347805,
      "compression_loss": 0.0,
      "distillation_loss": 0.2309635877609253,
      "epoch": 3.41,
      "learning_rate": 2.607052476747639e-05,
      "loss": 0.222,
      "step": 3588,
      "task_loss": 0.14136911928653717
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7810807549662145,
      "compression_loss": 0.0,
      "distillation_loss": 0.024008475244045258,
      "epoch": 3.41,
      "learning_rate": 2.6059879198237026e-05,
      "loss": 0.0232,
      "step": 3589,
      "task_loss": 0.016412295401096344
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7810924890438419,
      "compression_loss": 0.0,
      "distillation_loss": 0.012282849289476871,
      "epoch": 3.41,
      "learning_rate": 2.6049233436465498e-05,
      "loss": 0.0191,
      "step": 3590,
      "task_loss": 0.08071555197238922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7811042182686664,
      "compression_loss": 0.0,
      "distillation_loss": 0.01852232962846756,
      "epoch": 3.41,
      "learning_rate": 2.6038587484095673e-05,
      "loss": 0.0184,
      "step": 3591,
      "task_loss": 0.01706135831773281
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7811159426416916,
      "compression_loss": 0.0,
      "distillation_loss": 0.15857195854187012,
      "epoch": 3.41,
      "learning_rate": 2.6027941343061412e-05,
      "loss": 0.1539,
      "step": 3592,
      "task_loss": 0.11174768954515457
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7811276621639213,
      "compression_loss": 0.0,
      "distillation_loss": 0.12345419079065323,
      "epoch": 3.41,
      "learning_rate": 2.6017295015296665e-05,
      "loss": 0.122,
      "step": 3593,
      "task_loss": 0.10872073471546173
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.781139376836359,
      "compression_loss": 0.0,
      "distillation_loss": 0.19352051615715027,
      "epoch": 3.41,
      "learning_rate": 2.600664850273538e-05,
      "loss": 0.1846,
      "step": 3594,
      "task_loss": 0.10442230850458145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7811510866600085,
      "compression_loss": 0.0,
      "distillation_loss": 0.058610156178474426,
      "epoch": 3.41,
      "learning_rate": 2.599600180731155e-05,
      "loss": 0.0625,
      "step": 3595,
      "task_loss": 0.09727238863706589
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7811627916358737,
      "compression_loss": 0.0,
      "distillation_loss": 0.03841635584831238,
      "epoch": 3.42,
      "learning_rate": 2.598535493095919e-05,
      "loss": 0.0445,
      "step": 3596,
      "task_loss": 0.09931030124425888
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7811744917649579,
      "compression_loss": 0.0,
      "distillation_loss": 0.032603487372398376,
      "epoch": 3.42,
      "learning_rate": 2.5974707875612357e-05,
      "loss": 0.0327,
      "step": 3597,
      "task_loss": 0.0331152006983757
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.781186187048265,
      "compression_loss": 0.0,
      "distillation_loss": 0.07631140947341919,
      "epoch": 3.42,
      "learning_rate": 2.5964060643205153e-05,
      "loss": 0.0696,
      "step": 3598,
      "task_loss": 0.009496444836258888
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7811978774867987,
      "compression_loss": 0.0,
      "distillation_loss": 0.027815554291009903,
      "epoch": 3.42,
      "learning_rate": 2.5953413235671688e-05,
      "loss": 0.036,
      "step": 3599,
      "task_loss": 0.10932107269763947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7812095630815625,
      "compression_loss": 0.0,
      "distillation_loss": 0.07741496711969376,
      "epoch": 3.42,
      "learning_rate": 2.594276565494611e-05,
      "loss": 0.0841,
      "step": 3600,
      "task_loss": 0.14415237307548523
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7812212438335603,
      "compression_loss": 0.0,
      "distillation_loss": 0.09485390782356262,
      "epoch": 3.42,
      "learning_rate": 2.5932117902962616e-05,
      "loss": 0.0979,
      "step": 3601,
      "task_loss": 0.12560498714447021
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7812329197437957,
      "compression_loss": 0.0,
      "distillation_loss": 0.17144733667373657,
      "epoch": 3.42,
      "learning_rate": 2.5921469981655415e-05,
      "loss": 0.1648,
      "step": 3602,
      "task_loss": 0.10540945827960968
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7812445908132722,
      "compression_loss": 0.0,
      "distillation_loss": 0.11685089021921158,
      "epoch": 3.42,
      "learning_rate": 2.591082189295876e-05,
      "loss": 0.1339,
      "step": 3603,
      "task_loss": 0.28701621294021606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7812562570429938,
      "compression_loss": 0.0,
      "distillation_loss": 0.0379994660615921,
      "epoch": 3.42,
      "learning_rate": 2.590017363880691e-05,
      "loss": 0.0415,
      "step": 3604,
      "task_loss": 0.07254001498222351
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.781267918433964,
      "compression_loss": 0.0,
      "distillation_loss": 0.07432805001735687,
      "epoch": 3.42,
      "learning_rate": 2.5889525221134192e-05,
      "loss": 0.072,
      "step": 3605,
      "task_loss": 0.0511680468916893
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7812795749871866,
      "compression_loss": 0.0,
      "distillation_loss": 0.11450830101966858,
      "epoch": 3.42,
      "learning_rate": 2.5878876641874928e-05,
      "loss": 0.1202,
      "step": 3606,
      "task_loss": 0.1715685874223709
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.781291226703665,
      "compression_loss": 0.0,
      "distillation_loss": 0.0681910514831543,
      "epoch": 3.43,
      "learning_rate": 2.5868227902963493e-05,
      "loss": 0.064,
      "step": 3607,
      "task_loss": 0.026141859591007233
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7813028735844032,
      "compression_loss": 0.0,
      "distillation_loss": 0.056682735681533813,
      "epoch": 3.43,
      "learning_rate": 2.5857579006334282e-05,
      "loss": 0.0532,
      "step": 3608,
      "task_loss": 0.022059109061956406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7813145156304048,
      "compression_loss": 0.0,
      "distillation_loss": 0.08323369175195694,
      "epoch": 3.43,
      "learning_rate": 2.58469299539217e-05,
      "loss": 0.0813,
      "step": 3609,
      "task_loss": 0.06356573849916458
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7813261528426734,
      "compression_loss": 0.0,
      "distillation_loss": 0.015457273460924625,
      "epoch": 3.43,
      "learning_rate": 2.5836280747660225e-05,
      "loss": 0.0144,
      "step": 3610,
      "task_loss": 0.005144596099853516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7813377852222126,
      "compression_loss": 0.0,
      "distillation_loss": 0.05984696373343468,
      "epoch": 3.43,
      "learning_rate": 2.5825631389484323e-05,
      "loss": 0.0799,
      "step": 3611,
      "task_loss": 0.2608661949634552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7813494127700263,
      "compression_loss": 0.0,
      "distillation_loss": 0.22596868872642517,
      "epoch": 3.43,
      "learning_rate": 2.58149818813285e-05,
      "loss": 0.2201,
      "step": 3612,
      "task_loss": 0.16762208938598633
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7813610354871181,
      "compression_loss": 0.0,
      "distillation_loss": 0.09914842247962952,
      "epoch": 3.43,
      "learning_rate": 2.5804332225127294e-05,
      "loss": 0.101,
      "step": 3613,
      "task_loss": 0.117338627576828
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7813726533744917,
      "compression_loss": 0.0,
      "distillation_loss": 0.06026465445756912,
      "epoch": 3.43,
      "learning_rate": 2.579368242281527e-05,
      "loss": 0.0678,
      "step": 3614,
      "task_loss": 0.13608182966709137
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7813842664331506,
      "compression_loss": 0.0,
      "distillation_loss": 0.16431432962417603,
      "epoch": 3.43,
      "learning_rate": 2.5783032476327007e-05,
      "loss": 0.1575,
      "step": 3615,
      "task_loss": 0.09610229730606079
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7813958746640988,
      "compression_loss": 0.0,
      "distillation_loss": 0.22635792195796967,
      "epoch": 3.43,
      "learning_rate": 2.5772382387597128e-05,
      "loss": 0.2213,
      "step": 3616,
      "task_loss": 0.17567506432533264
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7814074780683398,
      "compression_loss": 0.0,
      "distillation_loss": 0.02461588941514492,
      "epoch": 3.43,
      "learning_rate": 2.5761732158560263e-05,
      "loss": 0.023,
      "step": 3617,
      "task_loss": 0.008299414068460464
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7814190766468772,
      "compression_loss": 0.0,
      "distillation_loss": 0.06911341845989227,
      "epoch": 3.44,
      "learning_rate": 2.5751081791151083e-05,
      "loss": 0.0656,
      "step": 3618,
      "task_loss": 0.03424249589443207
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7814306704007148,
      "compression_loss": 0.0,
      "distillation_loss": 0.21919073164463043,
      "epoch": 3.44,
      "learning_rate": 2.574043128730428e-05,
      "loss": 0.2192,
      "step": 3619,
      "task_loss": 0.2192375510931015
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7814422593308563,
      "compression_loss": 0.0,
      "distillation_loss": 0.052816398441791534,
      "epoch": 3.44,
      "learning_rate": 2.572978064895457e-05,
      "loss": 0.0481,
      "step": 3620,
      "task_loss": 0.006035482510924339
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7814538434383054,
      "compression_loss": 0.0,
      "distillation_loss": 0.19541379809379578,
      "epoch": 3.44,
      "learning_rate": 2.5719129878036686e-05,
      "loss": 0.1998,
      "step": 3621,
      "task_loss": 0.2397608757019043
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7814654227240656,
      "compression_loss": 0.0,
      "distillation_loss": 0.05258458852767944,
      "epoch": 3.44,
      "learning_rate": 2.5708478976485402e-05,
      "loss": 0.0484,
      "step": 3622,
      "task_loss": 0.010881522670388222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7814769971891408,
      "compression_loss": 0.0,
      "distillation_loss": 0.07996995747089386,
      "epoch": 3.44,
      "learning_rate": 2.569782794623549e-05,
      "loss": 0.0778,
      "step": 3623,
      "task_loss": 0.058584682643413544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7814885668345345,
      "compression_loss": 0.0,
      "distillation_loss": 0.26377159357070923,
      "epoch": 3.44,
      "learning_rate": 2.5687176789221784e-05,
      "loss": 0.2652,
      "step": 3624,
      "task_loss": 0.2783206105232239
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7815001316612505,
      "compression_loss": 0.0,
      "distillation_loss": 0.1098802238702774,
      "epoch": 3.44,
      "learning_rate": 2.5676525507379097e-05,
      "loss": 0.1033,
      "step": 3625,
      "task_loss": 0.04377390444278717
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7815116916702924,
      "compression_loss": 0.0,
      "distillation_loss": 0.19648560881614685,
      "epoch": 3.44,
      "learning_rate": 2.566587410264229e-05,
      "loss": 0.1934,
      "step": 3626,
      "task_loss": 0.16522561013698578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7815232468626641,
      "compression_loss": 0.0,
      "distillation_loss": 0.09389695525169373,
      "epoch": 3.44,
      "learning_rate": 2.565522257694625e-05,
      "loss": 0.0902,
      "step": 3627,
      "task_loss": 0.05693268030881882
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.781534797239369,
      "compression_loss": 0.0,
      "distillation_loss": 0.053470179438591,
      "epoch": 3.45,
      "learning_rate": 2.5644570932225874e-05,
      "loss": 0.0505,
      "step": 3628,
      "task_loss": 0.023593464866280556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.781546342801411,
      "compression_loss": 0.0,
      "distillation_loss": 0.13491879403591156,
      "epoch": 3.45,
      "learning_rate": 2.5633919170416087e-05,
      "loss": 0.1316,
      "step": 3629,
      "task_loss": 0.10193748027086258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7815578835497936,
      "compression_loss": 0.0,
      "distillation_loss": 0.0393562838435173,
      "epoch": 3.45,
      "learning_rate": 2.5623267293451826e-05,
      "loss": 0.0364,
      "step": 3630,
      "task_loss": 0.009963281452655792
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7815694194855205,
      "compression_loss": 0.0,
      "distillation_loss": 0.09000158309936523,
      "epoch": 3.45,
      "learning_rate": 2.5612615303268062e-05,
      "loss": 0.0853,
      "step": 3631,
      "task_loss": 0.04277219995856285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7815809506095955,
      "compression_loss": 0.0,
      "distillation_loss": 0.04296325892210007,
      "epoch": 3.45,
      "learning_rate": 2.560196320179977e-05,
      "loss": 0.0837,
      "step": 3632,
      "task_loss": 0.4504657983779907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7815924769230224,
      "compression_loss": 0.0,
      "distillation_loss": 0.05088837444782257,
      "epoch": 3.45,
      "learning_rate": 2.559131099098197e-05,
      "loss": 0.0472,
      "step": 3633,
      "task_loss": 0.013780592009425163
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7816039984268045,
      "compression_loss": 0.0,
      "distillation_loss": 0.04055938124656677,
      "epoch": 3.45,
      "learning_rate": 2.558065867274968e-05,
      "loss": 0.0381,
      "step": 3634,
      "task_loss": 0.015710243955254555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7816155151219457,
      "compression_loss": 0.0,
      "distillation_loss": 0.21613720059394836,
      "epoch": 3.45,
      "learning_rate": 2.5570006249037943e-05,
      "loss": 0.2207,
      "step": 3635,
      "task_loss": 0.261584997177124
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7816270270094497,
      "compression_loss": 0.0,
      "distillation_loss": 0.17852503061294556,
      "epoch": 3.45,
      "learning_rate": 2.5559353721781832e-05,
      "loss": 0.1745,
      "step": 3636,
      "task_loss": 0.13826580345630646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7816385340903202,
      "compression_loss": 0.0,
      "distillation_loss": 0.019594872370362282,
      "epoch": 3.45,
      "learning_rate": 2.5548701092916415e-05,
      "loss": 0.0184,
      "step": 3637,
      "task_loss": 0.00783473439514637
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7816500363655609,
      "compression_loss": 0.0,
      "distillation_loss": 0.028669090941548347,
      "epoch": 3.45,
      "learning_rate": 2.5538048364376806e-05,
      "loss": 0.0316,
      "step": 3638,
      "task_loss": 0.05839303508400917
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7816615338361753,
      "compression_loss": 0.0,
      "distillation_loss": 0.039027225226163864,
      "epoch": 3.46,
      "learning_rate": 2.552739553809812e-05,
      "loss": 0.0409,
      "step": 3639,
      "task_loss": 0.05817575752735138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7816730265031673,
      "compression_loss": 0.0,
      "distillation_loss": 0.05956938490271568,
      "epoch": 3.46,
      "learning_rate": 2.5516742616015493e-05,
      "loss": 0.0601,
      "step": 3640,
      "task_loss": 0.06536316871643066
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7816845143675404,
      "compression_loss": 0.0,
      "distillation_loss": 0.19583433866500854,
      "epoch": 3.46,
      "learning_rate": 2.5506089600064086e-05,
      "loss": 0.2043,
      "step": 3641,
      "task_loss": 0.2809743881225586
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7816959974302984,
      "compression_loss": 0.0,
      "distillation_loss": 0.10645139217376709,
      "epoch": 3.46,
      "learning_rate": 2.549543649217906e-05,
      "loss": 0.1073,
      "step": 3642,
      "task_loss": 0.11531829088926315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7817074756924449,
      "compression_loss": 0.0,
      "distillation_loss": 0.13260212540626526,
      "epoch": 3.46,
      "learning_rate": 2.548478329429561e-05,
      "loss": 0.1288,
      "step": 3643,
      "task_loss": 0.09502564370632172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7817189491549836,
      "compression_loss": 0.0,
      "distillation_loss": 0.03851540386676788,
      "epoch": 3.46,
      "learning_rate": 2.5474130008348946e-05,
      "loss": 0.0381,
      "step": 3644,
      "task_loss": 0.034055422991514206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7817304178189184,
      "compression_loss": 0.0,
      "distillation_loss": 0.03768332302570343,
      "epoch": 3.46,
      "learning_rate": 2.5463476636274276e-05,
      "loss": 0.0514,
      "step": 3645,
      "task_loss": 0.17461322247982025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7817418816852526,
      "compression_loss": 0.0,
      "distillation_loss": 0.19072747230529785,
      "epoch": 3.46,
      "learning_rate": 2.5452823180006845e-05,
      "loss": 0.1848,
      "step": 3646,
      "task_loss": 0.13176460564136505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7817533407549903,
      "compression_loss": 0.0,
      "distillation_loss": 0.06055070832371712,
      "epoch": 3.46,
      "learning_rate": 2.5442169641481907e-05,
      "loss": 0.0671,
      "step": 3647,
      "task_loss": 0.1260763555765152
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7817647950291348,
      "compression_loss": 0.0,
      "distillation_loss": 0.03400256484746933,
      "epoch": 3.46,
      "learning_rate": 2.5431516022634715e-05,
      "loss": 0.0456,
      "step": 3648,
      "task_loss": 0.1496407389640808
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7817762445086899,
      "compression_loss": 0.0,
      "distillation_loss": 0.021507391706109047,
      "epoch": 3.47,
      "learning_rate": 2.5420862325400563e-05,
      "loss": 0.02,
      "step": 3649,
      "task_loss": 0.006038764491677284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7817876891946594,
      "compression_loss": 0.0,
      "distillation_loss": 0.02982695959508419,
      "epoch": 3.47,
      "learning_rate": 2.5410208551714742e-05,
      "loss": 0.0378,
      "step": 3650,
      "task_loss": 0.10967092961072922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7817991290880468,
      "compression_loss": 0.0,
      "distillation_loss": 0.08754716068506241,
      "epoch": 3.47,
      "learning_rate": 2.539955470351257e-05,
      "loss": 0.0896,
      "step": 3651,
      "task_loss": 0.1080557107925415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7818105641898561,
      "compression_loss": 0.0,
      "distillation_loss": 0.03534093499183655,
      "epoch": 3.47,
      "learning_rate": 2.5388900782729347e-05,
      "loss": 0.0326,
      "step": 3652,
      "task_loss": 0.007803870365023613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7818219945010906,
      "compression_loss": 0.0,
      "distillation_loss": 0.07137121260166168,
      "epoch": 3.47,
      "learning_rate": 2.5378246791300435e-05,
      "loss": 0.077,
      "step": 3653,
      "task_loss": 0.12779828906059265
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7818334200227542,
      "compression_loss": 0.0,
      "distillation_loss": 0.2230396568775177,
      "epoch": 3.47,
      "learning_rate": 2.536759273116117e-05,
      "loss": 0.2194,
      "step": 3654,
      "task_loss": 0.18703873455524445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7818448407558506,
      "compression_loss": 0.0,
      "distillation_loss": 0.13464270532131195,
      "epoch": 3.47,
      "learning_rate": 2.5356938604246916e-05,
      "loss": 0.1454,
      "step": 3655,
      "task_loss": 0.24241343140602112
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7818562567013833,
      "compression_loss": 0.0,
      "distillation_loss": 0.03593993932008743,
      "epoch": 3.47,
      "learning_rate": 2.534628441249305e-05,
      "loss": 0.0394,
      "step": 3656,
      "task_loss": 0.07036207616329193
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7818676678603561,
      "compression_loss": 0.0,
      "distillation_loss": 0.13949531316757202,
      "epoch": 3.47,
      "learning_rate": 2.5335630157834937e-05,
      "loss": 0.1425,
      "step": 3657,
      "task_loss": 0.16993454098701477
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7818790742337728,
      "compression_loss": 0.0,
      "distillation_loss": 0.08820469677448273,
      "epoch": 3.47,
      "learning_rate": 2.5324975842208004e-05,
      "loss": 0.0803,
      "step": 3658,
      "task_loss": 0.008724292740225792
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7818904758226368,
      "compression_loss": 0.0,
      "distillation_loss": 0.12831662595272064,
      "epoch": 3.47,
      "learning_rate": 2.5314321467547635e-05,
      "loss": 0.1308,
      "step": 3659,
      "task_loss": 0.15305723249912262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7819018726279521,
      "compression_loss": 0.0,
      "distillation_loss": 0.02787201665341854,
      "epoch": 3.48,
      "learning_rate": 2.5303667035789262e-05,
      "loss": 0.0258,
      "step": 3660,
      "task_loss": 0.007001947611570358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7819132646507222,
      "compression_loss": 0.0,
      "distillation_loss": 0.02681071124970913,
      "epoch": 3.48,
      "learning_rate": 2.5293012548868306e-05,
      "loss": 0.0394,
      "step": 3661,
      "task_loss": 0.15266357362270355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7819246518919507,
      "compression_loss": 0.0,
      "distillation_loss": 0.07564039528369904,
      "epoch": 3.48,
      "learning_rate": 2.5282358008720213e-05,
      "loss": 0.0799,
      "step": 3662,
      "task_loss": 0.11861392855644226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7819360343526415,
      "compression_loss": 0.0,
      "distillation_loss": 0.04754069447517395,
      "epoch": 3.48,
      "learning_rate": 2.5271703417280433e-05,
      "loss": 0.0612,
      "step": 3663,
      "task_loss": 0.1845139116048813
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7819474120337981,
      "compression_loss": 0.0,
      "distillation_loss": 0.01500766258686781,
      "epoch": 3.48,
      "learning_rate": 2.526104877648441e-05,
      "loss": 0.0139,
      "step": 3664,
      "task_loss": 0.00399116612970829
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7819587849364243,
      "compression_loss": 0.0,
      "distillation_loss": 0.06251572072505951,
      "epoch": 3.48,
      "learning_rate": 2.525039408826762e-05,
      "loss": 0.065,
      "step": 3665,
      "task_loss": 0.08762296289205551
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7819701530615237,
      "compression_loss": 0.0,
      "distillation_loss": 0.1165790855884552,
      "epoch": 3.48,
      "learning_rate": 2.523973935456554e-05,
      "loss": 0.1153,
      "step": 3666,
      "task_loss": 0.10344909876585007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7819815164101,
      "compression_loss": 0.0,
      "distillation_loss": 0.09375932067632675,
      "epoch": 3.48,
      "learning_rate": 2.522908457731366e-05,
      "loss": 0.0982,
      "step": 3667,
      "task_loss": 0.138652503490448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7819928749831571,
      "compression_loss": 0.0,
      "distillation_loss": 0.09869857132434845,
      "epoch": 3.48,
      "learning_rate": 2.5218429758447455e-05,
      "loss": 0.0972,
      "step": 3668,
      "task_loss": 0.08332294970750809
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7820042287816983,
      "compression_loss": 0.0,
      "distillation_loss": 0.07920647412538528,
      "epoch": 3.48,
      "learning_rate": 2.520777489990243e-05,
      "loss": 0.0822,
      "step": 3669,
      "task_loss": 0.10958139598369598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7820155778067275,
      "compression_loss": 0.0,
      "distillation_loss": 0.04178139567375183,
      "epoch": 3.49,
      "learning_rate": 2.5197120003614094e-05,
      "loss": 0.0431,
      "step": 3670,
      "task_loss": 0.054654479026794434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7820269220592484,
      "compression_loss": 0.0,
      "distillation_loss": 0.08496996760368347,
      "epoch": 3.49,
      "learning_rate": 2.518646507151796e-05,
      "loss": 0.0805,
      "step": 3671,
      "task_loss": 0.04056711122393608
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7820382615402646,
      "compression_loss": 0.0,
      "distillation_loss": 0.07371478527784348,
      "epoch": 3.49,
      "learning_rate": 2.517581010554956e-05,
      "loss": 0.0778,
      "step": 3672,
      "task_loss": 0.11423461884260178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7820495962507797,
      "compression_loss": 0.0,
      "distillation_loss": 0.04122939705848694,
      "epoch": 3.49,
      "learning_rate": 2.5165155107644394e-05,
      "loss": 0.0386,
      "step": 3673,
      "task_loss": 0.014970516785979271
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7820609261917976,
      "compression_loss": 0.0,
      "distillation_loss": 0.18131166696548462,
      "epoch": 3.49,
      "learning_rate": 2.515450007973801e-05,
      "loss": 0.1739,
      "step": 3674,
      "task_loss": 0.10724371671676636
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7820722513643219,
      "compression_loss": 0.0,
      "distillation_loss": 0.11633885651826859,
      "epoch": 3.49,
      "learning_rate": 2.5143845023765943e-05,
      "loss": 0.1146,
      "step": 3675,
      "task_loss": 0.09903288632631302
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7820835717693562,
      "compression_loss": 0.0,
      "distillation_loss": 0.15314285457134247,
      "epoch": 3.49,
      "learning_rate": 2.513318994166373e-05,
      "loss": 0.1471,
      "step": 3676,
      "task_loss": 0.09299644827842712
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7820948874079042,
      "compression_loss": 0.0,
      "distillation_loss": 0.05131184682250023,
      "epoch": 3.49,
      "learning_rate": 2.5122534835366934e-05,
      "loss": 0.0473,
      "step": 3677,
      "task_loss": 0.011473558843135834
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7821061982809697,
      "compression_loss": 0.0,
      "distillation_loss": 0.23348447680473328,
      "epoch": 3.49,
      "learning_rate": 2.5111879706811087e-05,
      "loss": 0.226,
      "step": 3678,
      "task_loss": 0.15873844921588898
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7821175043895563,
      "compression_loss": 0.0,
      "distillation_loss": 0.050928764045238495,
      "epoch": 3.49,
      "learning_rate": 2.5101224557931758e-05,
      "loss": 0.0542,
      "step": 3679,
      "task_loss": 0.08366743475198746
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7821288057346677,
      "compression_loss": 0.0,
      "distillation_loss": 0.1100388914346695,
      "epoch": 3.49,
      "learning_rate": 2.5090569390664492e-05,
      "loss": 0.1145,
      "step": 3680,
      "task_loss": 0.15434427559375763
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7821401023173075,
      "compression_loss": 0.0,
      "distillation_loss": 0.13357719779014587,
      "epoch": 3.5,
      "learning_rate": 2.5079914206944866e-05,
      "loss": 0.1348,
      "step": 3681,
      "task_loss": 0.14616169035434723
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7821513941384794,
      "compression_loss": 0.0,
      "distillation_loss": 0.10456112027168274,
      "epoch": 3.5,
      "learning_rate": 2.5069259008708446e-05,
      "loss": 0.1072,
      "step": 3682,
      "task_loss": 0.1304989755153656
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7821626811991873,
      "compression_loss": 0.0,
      "distillation_loss": 0.028667902573943138,
      "epoch": 3.5,
      "learning_rate": 2.5058603797890778e-05,
      "loss": 0.0353,
      "step": 3683,
      "task_loss": 0.094968281686306
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7821739635004347,
      "compression_loss": 0.0,
      "distillation_loss": 0.11443310976028442,
      "epoch": 3.5,
      "learning_rate": 2.504794857642746e-05,
      "loss": 0.1115,
      "step": 3684,
      "task_loss": 0.08492926508188248
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7821852410432252,
      "compression_loss": 0.0,
      "distillation_loss": 0.04449796676635742,
      "epoch": 3.5,
      "learning_rate": 2.5037293346254044e-05,
      "loss": 0.0453,
      "step": 3685,
      "task_loss": 0.05215233191847801
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7821965138285626,
      "compression_loss": 0.0,
      "distillation_loss": 0.08054036647081375,
      "epoch": 3.5,
      "learning_rate": 2.5026638109306118e-05,
      "loss": 0.0923,
      "step": 3686,
      "task_loss": 0.1976814568042755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7822077818574505,
      "compression_loss": 0.0,
      "distillation_loss": 0.03564410284161568,
      "epoch": 3.5,
      "learning_rate": 2.5015982867519245e-05,
      "loss": 0.0327,
      "step": 3687,
      "task_loss": 0.006579475477337837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7822190451308928,
      "compression_loss": 0.0,
      "distillation_loss": 0.032507769763469696,
      "epoch": 3.5,
      "learning_rate": 2.500532762282901e-05,
      "loss": 0.0497,
      "step": 3688,
      "task_loss": 0.20428074896335602
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7822303036498929,
      "compression_loss": 0.0,
      "distillation_loss": 0.03502486273646355,
      "epoch": 3.5,
      "learning_rate": 2.4994672377170988e-05,
      "loss": 0.0328,
      "step": 3689,
      "task_loss": 0.013175180181860924
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7822415574154546,
      "compression_loss": 0.0,
      "distillation_loss": 0.05492643266916275,
      "epoch": 3.5,
      "learning_rate": 2.4984017132480754e-05,
      "loss": 0.0587,
      "step": 3690,
      "task_loss": 0.0923444926738739
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7822528064285817,
      "compression_loss": 0.0,
      "distillation_loss": 0.022486487403512,
      "epoch": 3.51,
      "learning_rate": 2.4973361890693888e-05,
      "loss": 0.021,
      "step": 3691,
      "task_loss": 0.0075523629784584045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7822640506902777,
      "compression_loss": 0.0,
      "distillation_loss": 0.11572916805744171,
      "epoch": 3.51,
      "learning_rate": 2.496270665374596e-05,
      "loss": 0.1192,
      "step": 3692,
      "task_loss": 0.15077991783618927
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7822752902015464,
      "compression_loss": 0.0,
      "distillation_loss": 0.08084140717983246,
      "epoch": 3.51,
      "learning_rate": 2.4952051423572548e-05,
      "loss": 0.0787,
      "step": 3693,
      "task_loss": 0.058929864317178726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7822865249633913,
      "compression_loss": 0.0,
      "distillation_loss": 0.06542964279651642,
      "epoch": 3.51,
      "learning_rate": 2.494139620210923e-05,
      "loss": 0.0636,
      "step": 3694,
      "task_loss": 0.04673419147729874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7822977549768164,
      "compression_loss": 0.0,
      "distillation_loss": 0.0398731529712677,
      "epoch": 3.51,
      "learning_rate": 2.4930740991291567e-05,
      "loss": 0.0375,
      "step": 3695,
      "task_loss": 0.016480809077620506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7823089802428251,
      "compression_loss": 0.0,
      "distillation_loss": 0.02609540894627571,
      "epoch": 3.51,
      "learning_rate": 2.492008579305514e-05,
      "loss": 0.032,
      "step": 3696,
      "task_loss": 0.08469408750534058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7823202007624213,
      "compression_loss": 0.0,
      "distillation_loss": 0.07933405041694641,
      "epoch": 3.51,
      "learning_rate": 2.4909430609335517e-05,
      "loss": 0.0922,
      "step": 3697,
      "task_loss": 0.2080336958169937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7823314165366084,
      "compression_loss": 0.0,
      "distillation_loss": 0.08961853384971619,
      "epoch": 3.51,
      "learning_rate": 2.4898775442068248e-05,
      "loss": 0.0827,
      "step": 3698,
      "task_loss": 0.020808879286050797
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7823426275663903,
      "compression_loss": 0.0,
      "distillation_loss": 0.028536148369312286,
      "epoch": 3.51,
      "learning_rate": 2.4888120293188916e-05,
      "loss": 0.0301,
      "step": 3699,
      "task_loss": 0.04397343844175339
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7823538338527707,
      "compression_loss": 0.0,
      "distillation_loss": 0.030012015253305435,
      "epoch": 3.51,
      "learning_rate": 2.487746516463307e-05,
      "loss": 0.0367,
      "step": 3700,
      "task_loss": 0.096799835562706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7823650353967531,
      "compression_loss": 0.0,
      "distillation_loss": 0.03244200348854065,
      "epoch": 3.51,
      "learning_rate": 2.486681005833627e-05,
      "loss": 0.0298,
      "step": 3701,
      "task_loss": 0.005791664123535156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7823762321993413,
      "compression_loss": 0.0,
      "distillation_loss": 0.19793766736984253,
      "epoch": 3.52,
      "learning_rate": 2.4856154976234063e-05,
      "loss": 0.1875,
      "step": 3702,
      "task_loss": 0.09377109259366989
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7823874242615391,
      "compression_loss": 0.0,
      "distillation_loss": 0.15221963822841644,
      "epoch": 3.52,
      "learning_rate": 2.4845499920261993e-05,
      "loss": 0.1428,
      "step": 3703,
      "task_loss": 0.05772307515144348
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.78239861158435,
      "compression_loss": 0.0,
      "distillation_loss": 0.031839657574892044,
      "epoch": 3.52,
      "learning_rate": 2.4834844892355615e-05,
      "loss": 0.0292,
      "step": 3704,
      "task_loss": 0.005376823246479034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7824097941687777,
      "compression_loss": 0.0,
      "distillation_loss": 0.03910262882709503,
      "epoch": 3.52,
      "learning_rate": 2.4824189894450453e-05,
      "loss": 0.0486,
      "step": 3705,
      "task_loss": 0.1344204545021057
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7824209720158258,
      "compression_loss": 0.0,
      "distillation_loss": 0.14387640357017517,
      "epoch": 3.52,
      "learning_rate": 2.4813534928482038e-05,
      "loss": 0.1372,
      "step": 3706,
      "task_loss": 0.07675487548112869
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7824321451264983,
      "compression_loss": 0.0,
      "distillation_loss": 0.05041220039129257,
      "epoch": 3.52,
      "learning_rate": 2.4802879996385905e-05,
      "loss": 0.0566,
      "step": 3707,
      "task_loss": 0.11191262304782867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7824433135017986,
      "compression_loss": 0.0,
      "distillation_loss": 0.1696065366268158,
      "epoch": 3.52,
      "learning_rate": 2.4792225100097578e-05,
      "loss": 0.161,
      "step": 3708,
      "task_loss": 0.08338792622089386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7824544771427305,
      "compression_loss": 0.0,
      "distillation_loss": 0.035738278180360794,
      "epoch": 3.52,
      "learning_rate": 2.4781570241552554e-05,
      "loss": 0.0337,
      "step": 3709,
      "task_loss": 0.015493502840399742
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7824656360502976,
      "compression_loss": 0.0,
      "distillation_loss": 0.18329283595085144,
      "epoch": 3.52,
      "learning_rate": 2.477091542268635e-05,
      "loss": 0.184,
      "step": 3710,
      "task_loss": 0.18989968299865723
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7824767902255038,
      "compression_loss": 0.0,
      "distillation_loss": 0.016138188540935516,
      "epoch": 3.52,
      "learning_rate": 2.4760260645434462e-05,
      "loss": 0.0148,
      "step": 3711,
      "task_loss": 0.0031305551528930664
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7824879396693524,
      "compression_loss": 0.0,
      "distillation_loss": 0.18617260456085205,
      "epoch": 3.53,
      "learning_rate": 2.474960591173238e-05,
      "loss": 0.1785,
      "step": 3712,
      "task_loss": 0.10942824184894562
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7824990843828474,
      "compression_loss": 0.0,
      "distillation_loss": 0.0715465247631073,
      "epoch": 3.53,
      "learning_rate": 2.47389512235156e-05,
      "loss": 0.0795,
      "step": 3713,
      "task_loss": 0.15127640962600708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7825102243669924,
      "compression_loss": 0.0,
      "distillation_loss": 0.04901812970638275,
      "epoch": 3.53,
      "learning_rate": 2.472829658271958e-05,
      "loss": 0.0451,
      "step": 3714,
      "task_loss": 0.01026928424835205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7825213596227909,
      "compression_loss": 0.0,
      "distillation_loss": 0.03361319750547409,
      "epoch": 3.53,
      "learning_rate": 2.4717641991279786e-05,
      "loss": 0.031,
      "step": 3715,
      "task_loss": 0.007931182160973549
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7825324901512469,
      "compression_loss": 0.0,
      "distillation_loss": 0.01693807728588581,
      "epoch": 3.53,
      "learning_rate": 2.4706987451131693e-05,
      "loss": 0.0202,
      "step": 3716,
      "task_loss": 0.04959682375192642
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7825436159533639,
      "compression_loss": 0.0,
      "distillation_loss": 0.04695942997932434,
      "epoch": 3.53,
      "learning_rate": 2.4696332964210743e-05,
      "loss": 0.0442,
      "step": 3717,
      "task_loss": 0.01931283064186573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7825547370301456,
      "compression_loss": 0.0,
      "distillation_loss": 0.04885145276784897,
      "epoch": 3.53,
      "learning_rate": 2.468567853245237e-05,
      "loss": 0.0489,
      "step": 3718,
      "task_loss": 0.049653246998786926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7825658533825957,
      "compression_loss": 0.0,
      "distillation_loss": 0.03596015274524689,
      "epoch": 3.53,
      "learning_rate": 2.4675024157792005e-05,
      "loss": 0.0335,
      "step": 3719,
      "task_loss": 0.011791346594691277
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7825769650117178,
      "compression_loss": 0.0,
      "distillation_loss": 0.08147747069597244,
      "epoch": 3.53,
      "learning_rate": 2.4664369842165068e-05,
      "loss": 0.0833,
      "step": 3720,
      "task_loss": 0.0997631698846817
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7825880719185156,
      "compression_loss": 0.0,
      "distillation_loss": 0.0436641164124012,
      "epoch": 3.53,
      "learning_rate": 2.4653715587506963e-05,
      "loss": 0.0399,
      "step": 3721,
      "task_loss": 0.006380394101142883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.782599174103993,
      "compression_loss": 0.0,
      "distillation_loss": 0.0826558843255043,
      "epoch": 3.53,
      "learning_rate": 2.4643061395753093e-05,
      "loss": 0.0923,
      "step": 3722,
      "task_loss": 0.178600013256073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7826102715691534,
      "compression_loss": 0.0,
      "distillation_loss": 0.07361429929733276,
      "epoch": 3.54,
      "learning_rate": 2.463240726883884e-05,
      "loss": 0.0684,
      "step": 3723,
      "task_loss": 0.021939026191830635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7826213643150007,
      "compression_loss": 0.0,
      "distillation_loss": 0.08207493275403976,
      "epoch": 3.54,
      "learning_rate": 2.4621753208699567e-05,
      "loss": 0.0901,
      "step": 3724,
      "task_loss": 0.16210611164569855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7826324523425384,
      "compression_loss": 0.0,
      "distillation_loss": 0.058608826249837875,
      "epoch": 3.54,
      "learning_rate": 2.4611099217270652e-05,
      "loss": 0.0535,
      "step": 3725,
      "task_loss": 0.007720122113823891
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7826435356527703,
      "compression_loss": 0.0,
      "distillation_loss": 0.03315757215023041,
      "epoch": 3.54,
      "learning_rate": 2.4600445296487436e-05,
      "loss": 0.0306,
      "step": 3726,
      "task_loss": 0.007330842316150665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7826546142467001,
      "compression_loss": 0.0,
      "distillation_loss": 0.12233985215425491,
      "epoch": 3.54,
      "learning_rate": 2.4589791448285264e-05,
      "loss": 0.1266,
      "step": 3727,
      "task_loss": 0.16475452482700348
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7826656881253313,
      "compression_loss": 0.0,
      "distillation_loss": 0.05593999847769737,
      "epoch": 3.54,
      "learning_rate": 2.4579137674599443e-05,
      "loss": 0.0616,
      "step": 3728,
      "task_loss": 0.11271567642688751
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7826767572896678,
      "compression_loss": 0.0,
      "distillation_loss": 0.12087702751159668,
      "epoch": 3.54,
      "learning_rate": 2.456848397736529e-05,
      "loss": 0.1295,
      "step": 3729,
      "task_loss": 0.20751793682575226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7826878217407132,
      "compression_loss": 0.0,
      "distillation_loss": 0.060698773711919785,
      "epoch": 3.54,
      "learning_rate": 2.455783035851811e-05,
      "loss": 0.0597,
      "step": 3730,
      "task_loss": 0.05119254067540169
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7826988814794711,
      "compression_loss": 0.0,
      "distillation_loss": 0.07374706864356995,
      "epoch": 3.54,
      "learning_rate": 2.4547176819993164e-05,
      "loss": 0.082,
      "step": 3731,
      "task_loss": 0.15632230043411255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7827099365069453,
      "compression_loss": 0.0,
      "distillation_loss": 0.07239855825901031,
      "epoch": 3.54,
      "learning_rate": 2.4536523363725727e-05,
      "loss": 0.071,
      "step": 3732,
      "task_loss": 0.058301545679569244
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7827209868241394,
      "compression_loss": 0.0,
      "distillation_loss": 0.02140430361032486,
      "epoch": 3.55,
      "learning_rate": 2.4525869991651057e-05,
      "loss": 0.02,
      "step": 3733,
      "task_loss": 0.006873439997434616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7827320324320571,
      "compression_loss": 0.0,
      "distillation_loss": 0.020102519541978836,
      "epoch": 3.55,
      "learning_rate": 2.4515216705704395e-05,
      "loss": 0.0195,
      "step": 3734,
      "task_loss": 0.014333127066493034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7827430733317022,
      "compression_loss": 0.0,
      "distillation_loss": 0.055659808218479156,
      "epoch": 3.55,
      "learning_rate": 2.4504563507820942e-05,
      "loss": 0.0636,
      "step": 3735,
      "task_loss": 0.1351262629032135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7827541095240782,
      "compression_loss": 0.0,
      "distillation_loss": 0.04852107912302017,
      "epoch": 3.55,
      "learning_rate": 2.449391039993592e-05,
      "loss": 0.0525,
      "step": 3736,
      "task_loss": 0.08783454447984695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7827651410101889,
      "compression_loss": 0.0,
      "distillation_loss": 0.029902100563049316,
      "epoch": 3.55,
      "learning_rate": 2.4483257383984513e-05,
      "loss": 0.0296,
      "step": 3737,
      "task_loss": 0.026881104335188866
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.782776167791038,
      "compression_loss": 0.0,
      "distillation_loss": 0.11988551914691925,
      "epoch": 3.55,
      "learning_rate": 2.4472604461901887e-05,
      "loss": 0.1168,
      "step": 3738,
      "task_loss": 0.08916487544775009
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7827871898676291,
      "compression_loss": 0.0,
      "distillation_loss": 0.09784644842147827,
      "epoch": 3.55,
      "learning_rate": 2.44619516356232e-05,
      "loss": 0.0946,
      "step": 3739,
      "task_loss": 0.06569681316614151
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7827982072409658,
      "compression_loss": 0.0,
      "distillation_loss": 0.06321494281291962,
      "epoch": 3.55,
      "learning_rate": 2.4451298907083597e-05,
      "loss": 0.0693,
      "step": 3740,
      "task_loss": 0.1242145225405693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7828092199120521,
      "compression_loss": 0.0,
      "distillation_loss": 0.028960630297660828,
      "epoch": 3.55,
      "learning_rate": 2.4440646278218177e-05,
      "loss": 0.0275,
      "step": 3741,
      "task_loss": 0.013882100582122803
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7828202278818913,
      "compression_loss": 0.0,
      "distillation_loss": 0.04928376153111458,
      "epoch": 3.55,
      "learning_rate": 2.442999375096206e-05,
      "loss": 0.0484,
      "step": 3742,
      "task_loss": 0.040039315819740295
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7828312311514873,
      "compression_loss": 0.0,
      "distillation_loss": 0.08106836676597595,
      "epoch": 3.55,
      "learning_rate": 2.4419341327250323e-05,
      "loss": 0.0827,
      "step": 3743,
      "task_loss": 0.09725627303123474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7828422297218438,
      "compression_loss": 0.0,
      "distillation_loss": 0.26687347888946533,
      "epoch": 3.56,
      "learning_rate": 2.4408689009018037e-05,
      "loss": 0.2664,
      "step": 3744,
      "task_loss": 0.26216715574264526
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7828532235939644,
      "compression_loss": 0.0,
      "distillation_loss": 0.061114098876714706,
      "epoch": 3.56,
      "learning_rate": 2.4398036798200235e-05,
      "loss": 0.0888,
      "step": 3745,
      "task_loss": 0.3378845453262329
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7828642127688528,
      "compression_loss": 0.0,
      "distillation_loss": 0.10098116099834442,
      "epoch": 3.56,
      "learning_rate": 2.4387384696731947e-05,
      "loss": 0.1014,
      "step": 3746,
      "task_loss": 0.10469695925712585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7828751972475126,
      "compression_loss": 0.0,
      "distillation_loss": 0.17749673128128052,
      "epoch": 3.56,
      "learning_rate": 2.4376732706548183e-05,
      "loss": 0.187,
      "step": 3747,
      "task_loss": 0.2721644937992096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7828861770309476,
      "compression_loss": 0.0,
      "distillation_loss": 0.0985332578420639,
      "epoch": 3.56,
      "learning_rate": 2.4366080829583922e-05,
      "loss": 0.1061,
      "step": 3748,
      "task_loss": 0.17437314987182617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7828971521201615,
      "compression_loss": 0.0,
      "distillation_loss": 0.06623338907957077,
      "epoch": 3.56,
      "learning_rate": 2.4355429067774135e-05,
      "loss": 0.0673,
      "step": 3749,
      "task_loss": 0.07646562159061432
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7829081225161579,
      "compression_loss": 0.0,
      "distillation_loss": 0.05294762924313545,
      "epoch": 3.56,
      "learning_rate": 2.4344777423053748e-05,
      "loss": 0.0511,
      "step": 3750,
      "task_loss": 0.03410526365041733
    },
    {
      "epoch": 3.56,
      "eval_accuracy": 0.8853211009174312,
      "eval_loss": 0.439721018075943,
      "eval_runtime": 18.0919,
      "eval_samples_per_second": 48.198,
      "eval_steps_per_second": 6.025,
      "step": 3750
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7829190882199406,
      "compression_loss": 0.0,
      "distillation_loss": 0.035928741097450256,
      "epoch": 3.56,
      "learning_rate": 2.433412589735771e-05,
      "loss": 0.033,
      "step": 3751,
      "task_loss": 0.0063820406794548035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7829300492325132,
      "compression_loss": 0.0,
      "distillation_loss": 0.04816991835832596,
      "epoch": 3.56,
      "learning_rate": 2.4323474492620905e-05,
      "loss": 0.0522,
      "step": 3752,
      "task_loss": 0.08798433095216751
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7829410055548792,
      "compression_loss": 0.0,
      "distillation_loss": 0.05488145351409912,
      "epoch": 3.56,
      "learning_rate": 2.431282321077822e-05,
      "loss": 0.0528,
      "step": 3753,
      "task_loss": 0.033642224967479706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7829519571880427,
      "compression_loss": 0.0,
      "distillation_loss": 0.04462762922048569,
      "epoch": 3.57,
      "learning_rate": 2.4302172053764514e-05,
      "loss": 0.0415,
      "step": 3754,
      "task_loss": 0.012937184423208237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.782962904133007,
      "compression_loss": 0.0,
      "distillation_loss": 0.13158389925956726,
      "epoch": 3.57,
      "learning_rate": 2.4291521023514604e-05,
      "loss": 0.1315,
      "step": 3755,
      "task_loss": 0.13095977902412415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7829738463907759,
      "compression_loss": 0.0,
      "distillation_loss": 0.12265504896640778,
      "epoch": 3.57,
      "learning_rate": 2.4280870121963323e-05,
      "loss": 0.118,
      "step": 3756,
      "task_loss": 0.0760970339179039
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7829847839623532,
      "compression_loss": 0.0,
      "distillation_loss": 0.03240625560283661,
      "epoch": 3.57,
      "learning_rate": 2.4270219351045438e-05,
      "loss": 0.0415,
      "step": 3757,
      "task_loss": 0.12327875196933746
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7829957168487424,
      "compression_loss": 0.0,
      "distillation_loss": 0.033418092876672745,
      "epoch": 3.57,
      "learning_rate": 2.425956871269572e-05,
      "loss": 0.0388,
      "step": 3758,
      "task_loss": 0.08721967041492462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7830066450509473,
      "compression_loss": 0.0,
      "distillation_loss": 0.038820862770080566,
      "epoch": 3.57,
      "learning_rate": 2.4248918208848916e-05,
      "loss": 0.0417,
      "step": 3759,
      "task_loss": 0.06801855564117432
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7830175685699716,
      "compression_loss": 0.0,
      "distillation_loss": 0.06290020793676376,
      "epoch": 3.57,
      "learning_rate": 2.423826784143974e-05,
      "loss": 0.0586,
      "step": 3760,
      "task_loss": 0.019695976749062538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7830284874068189,
      "compression_loss": 0.0,
      "distillation_loss": 0.04252585023641586,
      "epoch": 3.57,
      "learning_rate": 2.422761761240288e-05,
      "loss": 0.0478,
      "step": 3761,
      "task_loss": 0.09514269232749939
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.783039401562493,
      "compression_loss": 0.0,
      "distillation_loss": 0.22789345681667328,
      "epoch": 3.57,
      "learning_rate": 2.4216967523673e-05,
      "loss": 0.22,
      "step": 3762,
      "task_loss": 0.14860832691192627
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7830503110379974,
      "compression_loss": 0.0,
      "distillation_loss": 0.13879983127117157,
      "epoch": 3.57,
      "learning_rate": 2.420631757718474e-05,
      "loss": 0.1326,
      "step": 3763,
      "task_loss": 0.07688990235328674
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7830612158343359,
      "compression_loss": 0.0,
      "distillation_loss": 0.14175349473953247,
      "epoch": 3.57,
      "learning_rate": 2.419566777487271e-05,
      "loss": 0.138,
      "step": 3764,
      "task_loss": 0.1046019047498703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7830721159525121,
      "compression_loss": 0.0,
      "distillation_loss": 0.02354617603123188,
      "epoch": 3.58,
      "learning_rate": 2.4185018118671504e-05,
      "loss": 0.0316,
      "step": 3765,
      "task_loss": 0.1045549288392067
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7830830113935299,
      "compression_loss": 0.0,
      "distillation_loss": 0.23610442876815796,
      "epoch": 3.58,
      "learning_rate": 2.417436861051569e-05,
      "loss": 0.2247,
      "step": 3766,
      "task_loss": 0.12234364449977875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7830939021583927,
      "compression_loss": 0.0,
      "distillation_loss": 0.08728310465812683,
      "epoch": 3.58,
      "learning_rate": 2.4163719252339774e-05,
      "loss": 0.0861,
      "step": 3767,
      "task_loss": 0.0753721296787262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7831047882481044,
      "compression_loss": 0.0,
      "distillation_loss": 0.04650529474020004,
      "epoch": 3.58,
      "learning_rate": 2.41530700460783e-05,
      "loss": 0.0705,
      "step": 3768,
      "task_loss": 0.2862272262573242
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7831156696636685,
      "compression_loss": 0.0,
      "distillation_loss": 0.03932388499379158,
      "epoch": 3.58,
      "learning_rate": 2.4142420993665727e-05,
      "loss": 0.0426,
      "step": 3769,
      "task_loss": 0.072311170399189
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7831265464060888,
      "compression_loss": 0.0,
      "distillation_loss": 0.09386499226093292,
      "epoch": 3.58,
      "learning_rate": 2.4131772097036516e-05,
      "loss": 0.0865,
      "step": 3770,
      "task_loss": 0.0198683962225914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.783137418476369,
      "compression_loss": 0.0,
      "distillation_loss": 0.09301801770925522,
      "epoch": 3.58,
      "learning_rate": 2.4121123358125078e-05,
      "loss": 0.1046,
      "step": 3771,
      "task_loss": 0.20893818140029907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7831482858755127,
      "compression_loss": 0.0,
      "distillation_loss": 0.15907953679561615,
      "epoch": 3.58,
      "learning_rate": 2.4110474778865817e-05,
      "loss": 0.1651,
      "step": 3772,
      "task_loss": 0.21910780668258667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7831591486045236,
      "compression_loss": 0.0,
      "distillation_loss": 0.050729766488075256,
      "epoch": 3.58,
      "learning_rate": 2.4099826361193098e-05,
      "loss": 0.0566,
      "step": 3773,
      "task_loss": 0.10932128876447678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7831700066644055,
      "compression_loss": 0.0,
      "distillation_loss": 0.0897529125213623,
      "epoch": 3.58,
      "learning_rate": 2.4089178107041254e-05,
      "loss": 0.0848,
      "step": 3774,
      "task_loss": 0.039870765060186386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7831808600561619,
      "compression_loss": 0.0,
      "distillation_loss": 0.1213412880897522,
      "epoch": 3.58,
      "learning_rate": 2.407853001834459e-05,
      "loss": 0.1197,
      "step": 3775,
      "task_loss": 0.10536501556634903
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7831917087807966,
      "compression_loss": 0.0,
      "distillation_loss": 0.038229409605264664,
      "epoch": 3.59,
      "learning_rate": 2.4067882097037383e-05,
      "loss": 0.0357,
      "step": 3776,
      "task_loss": 0.012635331600904465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7832025528393133,
      "compression_loss": 0.0,
      "distillation_loss": 0.2114480882883072,
      "epoch": 3.59,
      "learning_rate": 2.4057234345053894e-05,
      "loss": 0.204,
      "step": 3777,
      "task_loss": 0.13707035779953003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7832133922327156,
      "compression_loss": 0.0,
      "distillation_loss": 0.03434763103723526,
      "epoch": 3.59,
      "learning_rate": 2.404658676432832e-05,
      "loss": 0.0492,
      "step": 3778,
      "task_loss": 0.18263056874275208
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7832242269620071,
      "compression_loss": 0.0,
      "distillation_loss": 0.024286724627017975,
      "epoch": 3.59,
      "learning_rate": 2.403593935679485e-05,
      "loss": 0.0226,
      "step": 3779,
      "task_loss": 0.0070180464535951614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7832350570281917,
      "compression_loss": 0.0,
      "distillation_loss": 0.05927498638629913,
      "epoch": 3.59,
      "learning_rate": 2.402529212438765e-05,
      "loss": 0.0679,
      "step": 3780,
      "task_loss": 0.1452222764492035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.783245882432273,
      "compression_loss": 0.0,
      "distillation_loss": 0.03792211413383484,
      "epoch": 3.59,
      "learning_rate": 2.401464506904082e-05,
      "loss": 0.0416,
      "step": 3781,
      "task_loss": 0.07421617209911346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7832567031752546,
      "compression_loss": 0.0,
      "distillation_loss": 0.1570988893508911,
      "epoch": 3.59,
      "learning_rate": 2.400399819268846e-05,
      "loss": 0.1591,
      "step": 3782,
      "task_loss": 0.1775505691766739
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7832675192581403,
      "compression_loss": 0.0,
      "distillation_loss": 0.019645625725388527,
      "epoch": 3.59,
      "learning_rate": 2.399335149726463e-05,
      "loss": 0.0182,
      "step": 3783,
      "task_loss": 0.0054306890815496445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7832783306819336,
      "compression_loss": 0.0,
      "distillation_loss": 0.0738186463713646,
      "epoch": 3.59,
      "learning_rate": 2.3982704984703337e-05,
      "loss": 0.068,
      "step": 3784,
      "task_loss": 0.015383878722786903
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7832891374476385,
      "compression_loss": 0.0,
      "distillation_loss": 0.02696252055466175,
      "epoch": 3.59,
      "learning_rate": 2.3972058656938587e-05,
      "loss": 0.0331,
      "step": 3785,
      "task_loss": 0.08802430331707001
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7832999395562583,
      "compression_loss": 0.0,
      "distillation_loss": 0.05371105670928955,
      "epoch": 3.6,
      "learning_rate": 2.3961412515904336e-05,
      "loss": 0.0574,
      "step": 3786,
      "task_loss": 0.09069317579269409
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.783310737008797,
      "compression_loss": 0.0,
      "distillation_loss": 0.0783916711807251,
      "epoch": 3.6,
      "learning_rate": 2.3950766563534508e-05,
      "loss": 0.0778,
      "step": 3787,
      "task_loss": 0.07237330079078674
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7833215298062581,
      "compression_loss": 0.0,
      "distillation_loss": 0.18467172980308533,
      "epoch": 3.6,
      "learning_rate": 2.394012080176298e-05,
      "loss": 0.1865,
      "step": 3788,
      "task_loss": 0.20330506563186646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7833323179496454,
      "compression_loss": 0.0,
      "distillation_loss": 0.014792121946811676,
      "epoch": 3.6,
      "learning_rate": 2.3929475232523612e-05,
      "loss": 0.0138,
      "step": 3789,
      "task_loss": 0.004657162353396416
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7833431014399624,
      "compression_loss": 0.0,
      "distillation_loss": 0.023432932794094086,
      "epoch": 3.6,
      "learning_rate": 2.3918829857750233e-05,
      "loss": 0.0217,
      "step": 3790,
      "task_loss": 0.0064360033720731735
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.783353880278213,
      "compression_loss": 0.0,
      "distillation_loss": 0.023829486221075058,
      "epoch": 3.6,
      "learning_rate": 2.3908184679376608e-05,
      "loss": 0.0219,
      "step": 3791,
      "task_loss": 0.004044756293296814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7833646544654007,
      "compression_loss": 0.0,
      "distillation_loss": 0.10334857553243637,
      "epoch": 3.6,
      "learning_rate": 2.3897539699336503e-05,
      "loss": 0.1155,
      "step": 3792,
      "task_loss": 0.22499464452266693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7833754240025294,
      "compression_loss": 0.0,
      "distillation_loss": 0.020630966871976852,
      "epoch": 3.6,
      "learning_rate": 2.3886894919563603e-05,
      "loss": 0.0257,
      "step": 3793,
      "task_loss": 0.07151245325803757
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7833861888906025,
      "compression_loss": 0.0,
      "distillation_loss": 0.09565474092960358,
      "epoch": 3.6,
      "learning_rate": 2.387625034199162e-05,
      "loss": 0.0906,
      "step": 3794,
      "task_loss": 0.044928256422281265
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7833969491306239,
      "compression_loss": 0.0,
      "distillation_loss": 0.061953071504831314,
      "epoch": 3.6,
      "learning_rate": 2.3865605968554163e-05,
      "loss": 0.0614,
      "step": 3795,
      "task_loss": 0.056104809045791626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7834077047235971,
      "compression_loss": 0.0,
      "distillation_loss": 0.050460297614336014,
      "epoch": 3.6,
      "learning_rate": 2.385496180118485e-05,
      "loss": 0.0469,
      "step": 3796,
      "task_loss": 0.014742163941264153
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.783418455670526,
      "compression_loss": 0.0,
      "distillation_loss": 0.07291211187839508,
      "epoch": 3.61,
      "learning_rate": 2.3844317841817245e-05,
      "loss": 0.0822,
      "step": 3797,
      "task_loss": 0.1659778207540512
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7834292019724142,
      "compression_loss": 0.0,
      "distillation_loss": 0.21132490038871765,
      "epoch": 3.61,
      "learning_rate": 2.3833674092384864e-05,
      "loss": 0.2126,
      "step": 3798,
      "task_loss": 0.22443267703056335
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7834399436302653,
      "compression_loss": 0.0,
      "distillation_loss": 0.11052772402763367,
      "epoch": 3.61,
      "learning_rate": 2.3823030554821208e-05,
      "loss": 0.1048,
      "step": 3799,
      "task_loss": 0.053448185324668884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.783450680645083,
      "compression_loss": 0.0,
      "distillation_loss": 0.02982490137219429,
      "epoch": 3.61,
      "learning_rate": 2.3812387231059712e-05,
      "loss": 0.028,
      "step": 3800,
      "task_loss": 0.011723890900611877
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7834614130178711,
      "compression_loss": 0.0,
      "distillation_loss": 0.03554895892739296,
      "epoch": 3.61,
      "learning_rate": 2.38017441230338e-05,
      "loss": 0.0389,
      "step": 3801,
      "task_loss": 0.0686713308095932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7834721407496331,
      "compression_loss": 0.0,
      "distillation_loss": 0.03221399337053299,
      "epoch": 3.61,
      "learning_rate": 2.379110123267683e-05,
      "loss": 0.0301,
      "step": 3802,
      "task_loss": 0.011455537751317024
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7834828638413729,
      "compression_loss": 0.0,
      "distillation_loss": 0.09096793085336685,
      "epoch": 3.61,
      "learning_rate": 2.3780458561922147e-05,
      "loss": 0.0944,
      "step": 3803,
      "task_loss": 0.12568344175815582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.783493582294094,
      "compression_loss": 0.0,
      "distillation_loss": 0.034135833382606506,
      "epoch": 3.61,
      "learning_rate": 2.3769816112703047e-05,
      "loss": 0.0383,
      "step": 3804,
      "task_loss": 0.07552683353424072
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7835042961088001,
      "compression_loss": 0.0,
      "distillation_loss": 0.021872466430068016,
      "epoch": 3.61,
      "learning_rate": 2.375917388695277e-05,
      "loss": 0.0203,
      "step": 3805,
      "task_loss": 0.005825823172926903
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7835150052864951,
      "compression_loss": 0.0,
      "distillation_loss": 0.10276195406913757,
      "epoch": 3.61,
      "learning_rate": 2.3748531886604537e-05,
      "loss": 0.1085,
      "step": 3806,
      "task_loss": 0.15964293479919434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7835257098281824,
      "compression_loss": 0.0,
      "distillation_loss": 0.0308642890304327,
      "epoch": 3.62,
      "learning_rate": 2.3737890113591507e-05,
      "loss": 0.0476,
      "step": 3807,
      "task_loss": 0.1985035240650177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7835364097348657,
      "compression_loss": 0.0,
      "distillation_loss": 0.037669505923986435,
      "epoch": 3.62,
      "learning_rate": 2.372724856984682e-05,
      "loss": 0.0412,
      "step": 3808,
      "task_loss": 0.07280071079730988
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.783547105007549,
      "compression_loss": 0.0,
      "distillation_loss": 0.09298324584960938,
      "epoch": 3.62,
      "learning_rate": 2.3716607257303563e-05,
      "loss": 0.0889,
      "step": 3809,
      "task_loss": 0.051655180752277374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7835577956472356,
      "compression_loss": 0.0,
      "distillation_loss": 0.11761534959077835,
      "epoch": 3.62,
      "learning_rate": 2.370596617789476e-05,
      "loss": 0.1226,
      "step": 3810,
      "task_loss": 0.1676085889339447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7835684816549293,
      "compression_loss": 0.0,
      "distillation_loss": 0.11634726822376251,
      "epoch": 3.62,
      "learning_rate": 2.369532533355345e-05,
      "loss": 0.1063,
      "step": 3811,
      "task_loss": 0.01633966900408268
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.783579163031634,
      "compression_loss": 0.0,
      "distillation_loss": 0.09321273863315582,
      "epoch": 3.62,
      "learning_rate": 2.3684684726212574e-05,
      "loss": 0.0988,
      "step": 3812,
      "task_loss": 0.14873237907886505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7835898397783531,
      "compression_loss": 0.0,
      "distillation_loss": 0.0937819555401802,
      "epoch": 3.62,
      "learning_rate": 2.3674044357805058e-05,
      "loss": 0.0936,
      "step": 3813,
      "task_loss": 0.0917680636048317
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7836005118960904,
      "compression_loss": 0.0,
      "distillation_loss": 0.13557323813438416,
      "epoch": 3.62,
      "learning_rate": 2.3663404230263764e-05,
      "loss": 0.1304,
      "step": 3814,
      "task_loss": 0.08434079587459564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7836111793858495,
      "compression_loss": 0.0,
      "distillation_loss": 0.11180786788463593,
      "epoch": 3.62,
      "learning_rate": 2.3652764345521527e-05,
      "loss": 0.1161,
      "step": 3815,
      "task_loss": 0.15484829246997833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7836218422486342,
      "compression_loss": 0.0,
      "distillation_loss": 0.03923030570149422,
      "epoch": 3.62,
      "learning_rate": 2.364212470551114e-05,
      "loss": 0.0417,
      "step": 3816,
      "task_loss": 0.06417812407016754
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7836325004854481,
      "compression_loss": 0.0,
      "distillation_loss": 0.06536407768726349,
      "epoch": 3.62,
      "learning_rate": 2.3631485312165337e-05,
      "loss": 0.0612,
      "step": 3817,
      "task_loss": 0.023885080590844154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.783643154097295,
      "compression_loss": 0.0,
      "distillation_loss": 0.2747918665409088,
      "epoch": 3.63,
      "learning_rate": 2.3620846167416816e-05,
      "loss": 0.2686,
      "step": 3818,
      "task_loss": 0.21282097697257996
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7836538030851784,
      "compression_loss": 0.0,
      "distillation_loss": 0.03365989029407501,
      "epoch": 3.63,
      "learning_rate": 2.3610207273198224e-05,
      "loss": 0.0308,
      "step": 3819,
      "task_loss": 0.005554560571908951
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7836644474501022,
      "compression_loss": 0.0,
      "distillation_loss": 0.015433688648045063,
      "epoch": 3.63,
      "learning_rate": 2.359956863144219e-05,
      "loss": 0.0143,
      "step": 3820,
      "task_loss": 0.004542894661426544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7836750871930699,
      "compression_loss": 0.0,
      "distillation_loss": 0.08182269334793091,
      "epoch": 3.63,
      "learning_rate": 2.358893024408125e-05,
      "loss": 0.0829,
      "step": 3821,
      "task_loss": 0.09263917803764343
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7836857223150853,
      "compression_loss": 0.0,
      "distillation_loss": 0.010741151869297028,
      "epoch": 3.63,
      "learning_rate": 2.3578292113047927e-05,
      "loss": 0.0144,
      "step": 3822,
      "task_loss": 0.0472935251891613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7836963528171519,
      "compression_loss": 0.0,
      "distillation_loss": 0.10976468026638031,
      "epoch": 3.63,
      "learning_rate": 2.3567654240274694e-05,
      "loss": 0.1043,
      "step": 3823,
      "task_loss": 0.05536213517189026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7837069787002735,
      "compression_loss": 0.0,
      "distillation_loss": 0.03287689387798309,
      "epoch": 3.63,
      "learning_rate": 2.355701662769396e-05,
      "loss": 0.0514,
      "step": 3824,
      "task_loss": 0.21781839430332184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7837175999654539,
      "compression_loss": 0.0,
      "distillation_loss": 0.03763230890035629,
      "epoch": 3.63,
      "learning_rate": 2.3546379277238107e-05,
      "loss": 0.0392,
      "step": 3825,
      "task_loss": 0.05362531542778015
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7837282166136966,
      "compression_loss": 0.0,
      "distillation_loss": 0.05240745097398758,
      "epoch": 3.63,
      "learning_rate": 2.3535742190839464e-05,
      "loss": 0.0651,
      "step": 3826,
      "task_loss": 0.17899471521377563
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7837388286460054,
      "compression_loss": 0.0,
      "distillation_loss": 0.03421130031347275,
      "epoch": 3.63,
      "learning_rate": 2.3525105370430296e-05,
      "loss": 0.0388,
      "step": 3827,
      "task_loss": 0.07988087832927704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.783749436063384,
      "compression_loss": 0.0,
      "distillation_loss": 0.036636196076869965,
      "epoch": 3.64,
      "learning_rate": 2.351446881794284e-05,
      "loss": 0.0541,
      "step": 3828,
      "task_loss": 0.21091753244400024
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7837600388668359,
      "compression_loss": 0.0,
      "distillation_loss": 0.029964663088321686,
      "epoch": 3.64,
      "learning_rate": 2.3503832535309282e-05,
      "loss": 0.032,
      "step": 3829,
      "task_loss": 0.05079011991620064
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7837706370573649,
      "compression_loss": 0.0,
      "distillation_loss": 0.03737621754407883,
      "epoch": 3.64,
      "learning_rate": 2.3493196524461754e-05,
      "loss": 0.0481,
      "step": 3830,
      "task_loss": 0.14465513825416565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7837812306359748,
      "compression_loss": 0.0,
      "distillation_loss": 0.05270380899310112,
      "epoch": 3.64,
      "learning_rate": 2.3482560787332325e-05,
      "loss": 0.052,
      "step": 3831,
      "task_loss": 0.04599842429161072
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7837918196036691,
      "compression_loss": 0.0,
      "distillation_loss": 0.09582822024822235,
      "epoch": 3.64,
      "learning_rate": 2.3471925325853043e-05,
      "loss": 0.0926,
      "step": 3832,
      "task_loss": 0.06387031823396683
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7838024039614515,
      "compression_loss": 0.0,
      "distillation_loss": 0.04903242364525795,
      "epoch": 3.64,
      "learning_rate": 2.3461290141955886e-05,
      "loss": 0.0461,
      "step": 3833,
      "task_loss": 0.019668804481625557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7838129837103258,
      "compression_loss": 0.0,
      "distillation_loss": 0.03492574021220207,
      "epoch": 3.64,
      "learning_rate": 2.345065523757278e-05,
      "loss": 0.0341,
      "step": 3834,
      "task_loss": 0.02714421972632408
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7838235588512955,
      "compression_loss": 0.0,
      "distillation_loss": 0.08957400172948837,
      "epoch": 3.64,
      "learning_rate": 2.3440020614635618e-05,
      "loss": 0.0918,
      "step": 3835,
      "task_loss": 0.11195338517427444
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7838341293853646,
      "compression_loss": 0.0,
      "distillation_loss": 0.029562341049313545,
      "epoch": 3.64,
      "learning_rate": 2.342938627507621e-05,
      "loss": 0.0338,
      "step": 3836,
      "task_loss": 0.07171028107404709
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7838446953135364,
      "compression_loss": 0.0,
      "distillation_loss": 0.022509558126330376,
      "epoch": 3.64,
      "learning_rate": 2.3418752220826364e-05,
      "loss": 0.0208,
      "step": 3837,
      "task_loss": 0.004994381219148636
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7838552566368148,
      "compression_loss": 0.0,
      "distillation_loss": 0.16286085546016693,
      "epoch": 3.64,
      "learning_rate": 2.3408118453817786e-05,
      "loss": 0.1811,
      "step": 3838,
      "task_loss": 0.3450527787208557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7838658133562034,
      "compression_loss": 0.0,
      "distillation_loss": 0.09071379154920578,
      "epoch": 3.65,
      "learning_rate": 2.339748497598216e-05,
      "loss": 0.0975,
      "step": 3839,
      "task_loss": 0.1586238443851471
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.783876365472706,
      "compression_loss": 0.0,
      "distillation_loss": 0.06195370852947235,
      "epoch": 3.65,
      "learning_rate": 2.338685178925111e-05,
      "loss": 0.0588,
      "step": 3840,
      "task_loss": 0.030472123995423317
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7838869129873262,
      "compression_loss": 0.0,
      "distillation_loss": 0.12223900854587555,
      "epoch": 3.65,
      "learning_rate": 2.3376218895556196e-05,
      "loss": 0.1351,
      "step": 3841,
      "task_loss": 0.2507644295692444
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7838974559010676,
      "compression_loss": 0.0,
      "distillation_loss": 0.18645763397216797,
      "epoch": 3.65,
      "learning_rate": 2.3365586296828944e-05,
      "loss": 0.1835,
      "step": 3842,
      "task_loss": 0.15729686617851257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7839079942149341,
      "compression_loss": 0.0,
      "distillation_loss": 0.14159414172172546,
      "epoch": 3.65,
      "learning_rate": 2.335495399500081e-05,
      "loss": 0.1349,
      "step": 3843,
      "task_loss": 0.07418064028024673
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7839185279299291,
      "compression_loss": 0.0,
      "distillation_loss": 0.04566461965441704,
      "epoch": 3.65,
      "learning_rate": 2.3344321992003206e-05,
      "loss": 0.0428,
      "step": 3844,
      "task_loss": 0.017158547416329384
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7839290570470565,
      "compression_loss": 0.0,
      "distillation_loss": 0.15318387746810913,
      "epoch": 3.65,
      "learning_rate": 2.3333690289767477e-05,
      "loss": 0.1488,
      "step": 3845,
      "task_loss": 0.10943318903446198
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7839395815673199,
      "compression_loss": 0.0,
      "distillation_loss": 0.026680761948227882,
      "epoch": 3.65,
      "learning_rate": 2.3323058890224938e-05,
      "loss": 0.0248,
      "step": 3846,
      "task_loss": 0.008314013481140137
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7839501014917231,
      "compression_loss": 0.0,
      "distillation_loss": 0.03555554896593094,
      "epoch": 3.65,
      "learning_rate": 2.3312427795306835e-05,
      "loss": 0.0487,
      "step": 3847,
      "task_loss": 0.16655853390693665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7839606168212695,
      "compression_loss": 0.0,
      "distillation_loss": 0.03203408420085907,
      "epoch": 3.65,
      "learning_rate": 2.330179700694434e-05,
      "loss": 0.044,
      "step": 3848,
      "task_loss": 0.15183991193771362
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7839711275569631,
      "compression_loss": 0.0,
      "distillation_loss": 0.047503869980573654,
      "epoch": 3.66,
      "learning_rate": 2.329116652706861e-05,
      "loss": 0.0442,
      "step": 3849,
      "task_loss": 0.01422363892197609
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7839816336998073,
      "compression_loss": 0.0,
      "distillation_loss": 0.04679963365197182,
      "epoch": 3.66,
      "learning_rate": 2.3280536357610704e-05,
      "loss": 0.0571,
      "step": 3850,
      "task_loss": 0.1500503420829773
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7839921352508061,
      "compression_loss": 0.0,
      "distillation_loss": 0.020564064383506775,
      "epoch": 3.66,
      "learning_rate": 2.3269906500501647e-05,
      "loss": 0.0253,
      "step": 3851,
      "task_loss": 0.06842130422592163
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7840026322109629,
      "compression_loss": 0.0,
      "distillation_loss": 0.08808699995279312,
      "epoch": 3.66,
      "learning_rate": 2.325927695767241e-05,
      "loss": 0.0912,
      "step": 3852,
      "task_loss": 0.11908883601427078
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7840131245812815,
      "compression_loss": 0.0,
      "distillation_loss": 0.036189496517181396,
      "epoch": 3.66,
      "learning_rate": 2.3248647731053894e-05,
      "loss": 0.0553,
      "step": 3853,
      "task_loss": 0.22739224135875702
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7840236123627656,
      "compression_loss": 0.0,
      "distillation_loss": 0.05552596598863602,
      "epoch": 3.66,
      "learning_rate": 2.3238018822576947e-05,
      "loss": 0.0509,
      "step": 3854,
      "task_loss": 0.009532701224088669
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7840340955564189,
      "compression_loss": 0.0,
      "distillation_loss": 0.11583777517080307,
      "epoch": 3.66,
      "learning_rate": 2.3227390234172372e-05,
      "loss": 0.11,
      "step": 3855,
      "task_loss": 0.05766517296433449
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7840445741632449,
      "compression_loss": 0.0,
      "distillation_loss": 0.04743620380759239,
      "epoch": 3.66,
      "learning_rate": 2.32167619677709e-05,
      "loss": 0.0638,
      "step": 3856,
      "task_loss": 0.2115105539560318
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7840550481842475,
      "compression_loss": 0.0,
      "distillation_loss": 0.035333409905433655,
      "epoch": 3.66,
      "learning_rate": 2.3206134025303206e-05,
      "loss": 0.0433,
      "step": 3857,
      "task_loss": 0.11506626009941101
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7840655176204303,
      "compression_loss": 0.0,
      "distillation_loss": 0.06820636987686157,
      "epoch": 3.66,
      "learning_rate": 2.31955064086999e-05,
      "loss": 0.0815,
      "step": 3858,
      "task_loss": 0.20162644982337952
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.784075982472797,
      "compression_loss": 0.0,
      "distillation_loss": 0.11444197595119476,
      "epoch": 3.66,
      "learning_rate": 2.3184879119891552e-05,
      "loss": 0.1087,
      "step": 3859,
      "task_loss": 0.057123761624097824
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7840864427423512,
      "compression_loss": 0.0,
      "distillation_loss": 0.09845539927482605,
      "epoch": 3.67,
      "learning_rate": 2.317425216080865e-05,
      "loss": 0.111,
      "step": 3860,
      "task_loss": 0.22417478263378143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7840968984300967,
      "compression_loss": 0.0,
      "distillation_loss": 0.1071721538901329,
      "epoch": 3.67,
      "learning_rate": 2.3163625533381636e-05,
      "loss": 0.1044,
      "step": 3861,
      "task_loss": 0.0796838253736496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7841073495370371,
      "compression_loss": 0.0,
      "distillation_loss": 0.023740757256746292,
      "epoch": 3.67,
      "learning_rate": 2.315299923954088e-05,
      "loss": 0.0288,
      "step": 3862,
      "task_loss": 0.07385114580392838
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7841177960641761,
      "compression_loss": 0.0,
      "distillation_loss": 0.04698867350816727,
      "epoch": 3.67,
      "learning_rate": 2.314237328121672e-05,
      "loss": 0.0501,
      "step": 3863,
      "task_loss": 0.07777606695890427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7841282380125174,
      "compression_loss": 0.0,
      "distillation_loss": 0.03970280662178993,
      "epoch": 3.67,
      "learning_rate": 2.3131747660339394e-05,
      "loss": 0.0383,
      "step": 3864,
      "task_loss": 0.026047592982649803
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7841386753830647,
      "compression_loss": 0.0,
      "distillation_loss": 0.03670245781540871,
      "epoch": 3.67,
      "learning_rate": 2.3121122378839106e-05,
      "loss": 0.0432,
      "step": 3865,
      "task_loss": 0.10145539790391922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7841491081768216,
      "compression_loss": 0.0,
      "distillation_loss": 0.028605319559574127,
      "epoch": 3.67,
      "learning_rate": 2.3110497438645987e-05,
      "loss": 0.0363,
      "step": 3866,
      "task_loss": 0.10599753260612488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7841595363947919,
      "compression_loss": 0.0,
      "distillation_loss": 0.11348830163478851,
      "epoch": 3.67,
      "learning_rate": 2.3099872841690103e-05,
      "loss": 0.108,
      "step": 3867,
      "task_loss": 0.05827032029628754
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7841699600379791,
      "compression_loss": 0.0,
      "distillation_loss": 0.08587086945772171,
      "epoch": 3.67,
      "learning_rate": 2.308924858990147e-05,
      "loss": 0.0806,
      "step": 3868,
      "task_loss": 0.033285610377788544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7841803791073871,
      "compression_loss": 0.0,
      "distillation_loss": 0.08816444873809814,
      "epoch": 3.67,
      "learning_rate": 2.3078624685210042e-05,
      "loss": 0.0914,
      "step": 3869,
      "task_loss": 0.12038688361644745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7841907936040196,
      "compression_loss": 0.0,
      "distillation_loss": 0.13361284136772156,
      "epoch": 3.68,
      "learning_rate": 2.3068001129545686e-05,
      "loss": 0.1276,
      "step": 3870,
      "task_loss": 0.07371234893798828
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.78420120352888,
      "compression_loss": 0.0,
      "distillation_loss": 0.033696770668029785,
      "epoch": 3.68,
      "learning_rate": 2.305737792483822e-05,
      "loss": 0.0399,
      "step": 3871,
      "task_loss": 0.09580248594284058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7842116088829721,
      "compression_loss": 0.0,
      "distillation_loss": 0.023039262741804123,
      "epoch": 3.68,
      "learning_rate": 2.3046755073017416e-05,
      "loss": 0.0341,
      "step": 3872,
      "task_loss": 0.13404729962348938
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7842220096672998,
      "compression_loss": 0.0,
      "distillation_loss": 0.014469185844063759,
      "epoch": 3.68,
      "learning_rate": 2.303613257601297e-05,
      "loss": 0.0137,
      "step": 3873,
      "task_loss": 0.006705537438392639
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7842324058828666,
      "compression_loss": 0.0,
      "distillation_loss": 0.06789643317461014,
      "epoch": 3.68,
      "learning_rate": 2.302551043575449e-05,
      "loss": 0.0626,
      "step": 3874,
      "task_loss": 0.015251386910676956
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.784242797530676,
      "compression_loss": 0.0,
      "distillation_loss": 0.04347585514187813,
      "epoch": 3.68,
      "learning_rate": 2.301488865417155e-05,
      "loss": 0.0488,
      "step": 3875,
      "task_loss": 0.09680940955877304
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7842531846117321,
      "compression_loss": 0.0,
      "distillation_loss": 0.04327643662691116,
      "epoch": 3.68,
      "learning_rate": 2.3004267233193655e-05,
      "loss": 0.0394,
      "step": 3876,
      "task_loss": 0.004441501572728157
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7842635671270383,
      "compression_loss": 0.0,
      "distillation_loss": 0.04019487649202347,
      "epoch": 3.68,
      "learning_rate": 2.2993646174750217e-05,
      "loss": 0.0394,
      "step": 3877,
      "task_loss": 0.032011084258556366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7842739450775983,
      "compression_loss": 0.0,
      "distillation_loss": 0.024311494082212448,
      "epoch": 3.68,
      "learning_rate": 2.2983025480770627e-05,
      "loss": 0.026,
      "step": 3878,
      "task_loss": 0.040784791111946106
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7842843184644157,
      "compression_loss": 0.0,
      "distillation_loss": 0.1018083393573761,
      "epoch": 3.68,
      "learning_rate": 2.297240515318416e-05,
      "loss": 0.106,
      "step": 3879,
      "task_loss": 0.14343193173408508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7842946872884945,
      "compression_loss": 0.0,
      "distillation_loss": 0.12023551762104034,
      "epoch": 3.68,
      "learning_rate": 2.2961785193920058e-05,
      "loss": 0.1204,
      "step": 3880,
      "task_loss": 0.1214599758386612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7843050515508381,
      "compression_loss": 0.0,
      "distillation_loss": 0.16024965047836304,
      "epoch": 3.69,
      "learning_rate": 2.2951165604907497e-05,
      "loss": 0.1819,
      "step": 3881,
      "task_loss": 0.37690994143486023
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7843154112524503,
      "compression_loss": 0.0,
      "distillation_loss": 0.1132129579782486,
      "epoch": 3.69,
      "learning_rate": 2.2940546388075572e-05,
      "loss": 0.1066,
      "step": 3882,
      "task_loss": 0.04658997803926468
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7843257663943348,
      "compression_loss": 0.0,
      "distillation_loss": 0.08600252866744995,
      "epoch": 3.69,
      "learning_rate": 2.2929927545353323e-05,
      "loss": 0.0817,
      "step": 3883,
      "task_loss": 0.0432564802467823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7843361169774952,
      "compression_loss": 0.0,
      "distillation_loss": 0.06563156843185425,
      "epoch": 3.69,
      "learning_rate": 2.2919309078669697e-05,
      "loss": 0.0658,
      "step": 3884,
      "task_loss": 0.06688161194324493
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7843464630029351,
      "compression_loss": 0.0,
      "distillation_loss": 0.0378117635846138,
      "epoch": 3.69,
      "learning_rate": 2.2908690989953598e-05,
      "loss": 0.0374,
      "step": 3885,
      "task_loss": 0.03329796344041824
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7843568044716585,
      "compression_loss": 0.0,
      "distillation_loss": 0.08495815098285675,
      "epoch": 3.69,
      "learning_rate": 2.2898073281133853e-05,
      "loss": 0.0777,
      "step": 3886,
      "task_loss": 0.01194603368639946
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7843671413846687,
      "compression_loss": 0.0,
      "distillation_loss": 0.03463613986968994,
      "epoch": 3.69,
      "learning_rate": 2.2887455954139217e-05,
      "loss": 0.0355,
      "step": 3887,
      "task_loss": 0.04286995902657509
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7843774737429696,
      "compression_loss": 0.0,
      "distillation_loss": 0.05026715248823166,
      "epoch": 3.69,
      "learning_rate": 2.2876839010898377e-05,
      "loss": 0.0459,
      "step": 3888,
      "task_loss": 0.00638798251748085
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7843878015475649,
      "compression_loss": 0.0,
      "distillation_loss": 0.062215156853199005,
      "epoch": 3.69,
      "learning_rate": 2.286622245333996e-05,
      "loss": 0.0596,
      "step": 3889,
      "task_loss": 0.03652032092213631
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7843981247994581,
      "compression_loss": 0.0,
      "distillation_loss": 0.041499149054288864,
      "epoch": 3.69,
      "learning_rate": 2.2855606283392516e-05,
      "loss": 0.0464,
      "step": 3890,
      "task_loss": 0.09033425897359848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7844084434996532,
      "compression_loss": 0.0,
      "distillation_loss": 0.04093802720308304,
      "epoch": 3.7,
      "learning_rate": 2.2844990502984513e-05,
      "loss": 0.0507,
      "step": 3891,
      "task_loss": 0.13819225132465363
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7844187576491536,
      "compression_loss": 0.0,
      "distillation_loss": 0.09159182012081146,
      "epoch": 3.7,
      "learning_rate": 2.2834375114044375e-05,
      "loss": 0.0887,
      "step": 3892,
      "task_loss": 0.06316046416759491
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7844290672489631,
      "compression_loss": 0.0,
      "distillation_loss": 0.15046623349189758,
      "epoch": 3.7,
      "learning_rate": 2.2823760118500415e-05,
      "loss": 0.1502,
      "step": 3893,
      "task_loss": 0.14759182929992676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7844393723000853,
      "compression_loss": 0.0,
      "distillation_loss": 0.018377002328634262,
      "epoch": 3.7,
      "learning_rate": 2.2813145518280914e-05,
      "loss": 0.0169,
      "step": 3894,
      "task_loss": 0.003946490585803986
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.784449672803524,
      "compression_loss": 0.0,
      "distillation_loss": 0.09543585032224655,
      "epoch": 3.7,
      "learning_rate": 2.2802531315314065e-05,
      "loss": 0.101,
      "step": 3895,
      "task_loss": 0.15144692361354828
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7844599687602828,
      "compression_loss": 0.0,
      "distillation_loss": 0.17009879648685455,
      "epoch": 3.7,
      "learning_rate": 2.279191751152798e-05,
      "loss": 0.1709,
      "step": 3896,
      "task_loss": 0.17769216001033783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7844702601713653,
      "compression_loss": 0.0,
      "distillation_loss": 0.11712504923343658,
      "epoch": 3.7,
      "learning_rate": 2.2781304108850706e-05,
      "loss": 0.1132,
      "step": 3897,
      "task_loss": 0.07775254547595978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7844805470377754,
      "compression_loss": 0.0,
      "distillation_loss": 0.04855842888355255,
      "epoch": 3.7,
      "learning_rate": 2.2770691109210235e-05,
      "loss": 0.0595,
      "step": 3898,
      "task_loss": 0.1582665592432022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7844908293605166,
      "compression_loss": 0.0,
      "distillation_loss": 0.07292292267084122,
      "epoch": 3.7,
      "learning_rate": 2.2760078514534462e-05,
      "loss": 0.0764,
      "step": 3899,
      "task_loss": 0.10792665183544159
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7845011071405927,
      "compression_loss": 0.0,
      "distillation_loss": 0.10464300960302353,
      "epoch": 3.7,
      "learning_rate": 2.2749466326751213e-05,
      "loss": 0.1044,
      "step": 3900,
      "task_loss": 0.10204927623271942
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7845113803790073,
      "compression_loss": 0.0,
      "distillation_loss": 0.03131140023469925,
      "epoch": 3.7,
      "learning_rate": 2.273885454778824e-05,
      "loss": 0.0288,
      "step": 3901,
      "task_loss": 0.0059508830308914185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7845216490767641,
      "compression_loss": 0.0,
      "distillation_loss": 0.050689488649368286,
      "epoch": 3.71,
      "learning_rate": 2.272824317957324e-05,
      "loss": 0.0619,
      "step": 3902,
      "task_loss": 0.16312363743782043
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7845319132348668,
      "compression_loss": 0.0,
      "distillation_loss": 0.11499884724617004,
      "epoch": 3.71,
      "learning_rate": 2.2717632224033796e-05,
      "loss": 0.1087,
      "step": 3903,
      "task_loss": 0.051606349647045135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7845421728543192,
      "compression_loss": 0.0,
      "distillation_loss": 0.15508276224136353,
      "epoch": 3.71,
      "learning_rate": 2.2707021683097454e-05,
      "loss": 0.1501,
      "step": 3904,
      "task_loss": 0.10518581420183182
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7845524279361247,
      "compression_loss": 0.0,
      "distillation_loss": 0.0731472373008728,
      "epoch": 3.71,
      "learning_rate": 2.2696411558691672e-05,
      "loss": 0.072,
      "step": 3905,
      "task_loss": 0.06154020130634308
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7845626784812871,
      "compression_loss": 0.0,
      "distillation_loss": 0.08605273813009262,
      "epoch": 3.71,
      "learning_rate": 2.268580185274381e-05,
      "loss": 0.0926,
      "step": 3906,
      "task_loss": 0.15165117383003235
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7845729244908103,
      "compression_loss": 0.0,
      "distillation_loss": 0.020624032244086266,
      "epoch": 3.71,
      "learning_rate": 2.2675192567181197e-05,
      "loss": 0.019,
      "step": 3907,
      "task_loss": 0.004072193056344986
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7845831659656977,
      "compression_loss": 0.0,
      "distillation_loss": 0.09648367762565613,
      "epoch": 3.71,
      "learning_rate": 2.2664583703931047e-05,
      "loss": 0.0974,
      "step": 3908,
      "task_loss": 0.10594139993190765
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.784593402906953,
      "compression_loss": 0.0,
      "distillation_loss": 0.03631186485290527,
      "epoch": 3.71,
      "learning_rate": 2.265397526492052e-05,
      "loss": 0.0487,
      "step": 3909,
      "task_loss": 0.16029351949691772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7846036353155801,
      "compression_loss": 0.0,
      "distillation_loss": 0.10707633942365646,
      "epoch": 3.71,
      "learning_rate": 2.2643367252076684e-05,
      "loss": 0.1087,
      "step": 3910,
      "task_loss": 0.12293732911348343
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7846138631925825,
      "compression_loss": 0.0,
      "distillation_loss": 0.027988499030470848,
      "epoch": 3.71,
      "learning_rate": 2.263275966732653e-05,
      "loss": 0.0256,
      "step": 3911,
      "task_loss": 0.00423845648765564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.784624086538964,
      "compression_loss": 0.0,
      "distillation_loss": 0.06167607381939888,
      "epoch": 3.72,
      "learning_rate": 2.262215251259699e-05,
      "loss": 0.0595,
      "step": 3912,
      "task_loss": 0.04003376513719559
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7846343053557281,
      "compression_loss": 0.0,
      "distillation_loss": 0.10655433684587479,
      "epoch": 3.72,
      "learning_rate": 2.2611545789814898e-05,
      "loss": 0.111,
      "step": 3913,
      "task_loss": 0.15064670145511627
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7846445196438786,
      "compression_loss": 0.0,
      "distillation_loss": 0.04435639828443527,
      "epoch": 3.72,
      "learning_rate": 2.2600939500907007e-05,
      "loss": 0.0445,
      "step": 3914,
      "task_loss": 0.04605961963534355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7846547294044193,
      "compression_loss": 0.0,
      "distillation_loss": 0.08209598064422607,
      "epoch": 3.72,
      "learning_rate": 2.2590333647800013e-05,
      "loss": 0.0942,
      "step": 3915,
      "task_loss": 0.2029554843902588
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7846649346383536,
      "compression_loss": 0.0,
      "distillation_loss": 0.03713390976190567,
      "epoch": 3.72,
      "learning_rate": 2.2579728232420525e-05,
      "loss": 0.0419,
      "step": 3916,
      "task_loss": 0.08507491648197174
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7846751353466855,
      "compression_loss": 0.0,
      "distillation_loss": 0.08504173159599304,
      "epoch": 3.72,
      "learning_rate": 2.2569123256695056e-05,
      "loss": 0.0915,
      "step": 3917,
      "task_loss": 0.1496903896331787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7846853315304185,
      "compression_loss": 0.0,
      "distillation_loss": 0.06336773186922073,
      "epoch": 3.72,
      "learning_rate": 2.2558518722550048e-05,
      "loss": 0.0811,
      "step": 3918,
      "task_loss": 0.2407415211200714
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7846955231905562,
      "compression_loss": 0.0,
      "distillation_loss": 0.08145183324813843,
      "epoch": 3.72,
      "learning_rate": 2.2547914631911884e-05,
      "loss": 0.0866,
      "step": 3919,
      "task_loss": 0.1333352029323578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7847057103281025,
      "compression_loss": 0.0,
      "distillation_loss": 0.10730397701263428,
      "epoch": 3.72,
      "learning_rate": 2.2537310986706826e-05,
      "loss": 0.1128,
      "step": 3920,
      "task_loss": 0.16203323006629944
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.784715892944061,
      "compression_loss": 0.0,
      "distillation_loss": 0.041547950357198715,
      "epoch": 3.72,
      "learning_rate": 2.2526707788861098e-05,
      "loss": 0.0386,
      "step": 3921,
      "task_loss": 0.011634528636932373
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7847260710394351,
      "compression_loss": 0.0,
      "distillation_loss": 0.06141505762934685,
      "epoch": 3.72,
      "learning_rate": 2.2516105040300804e-05,
      "loss": 0.0588,
      "step": 3922,
      "task_loss": 0.03478226065635681
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.784736244615229,
      "compression_loss": 0.0,
      "distillation_loss": 0.06806781888008118,
      "epoch": 3.73,
      "learning_rate": 2.250550274295199e-05,
      "loss": 0.0689,
      "step": 3923,
      "task_loss": 0.07671485841274261
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.784746413672446,
      "compression_loss": 0.0,
      "distillation_loss": 0.05792640894651413,
      "epoch": 3.73,
      "learning_rate": 2.2494900898740616e-05,
      "loss": 0.0609,
      "step": 3924,
      "task_loss": 0.08743541687726974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.78475657821209,
      "compression_loss": 0.0,
      "distillation_loss": 0.04159487783908844,
      "epoch": 3.73,
      "learning_rate": 2.2484299509592563e-05,
      "loss": 0.0539,
      "step": 3925,
      "task_loss": 0.16499102115631104
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7847667382351645,
      "compression_loss": 0.0,
      "distillation_loss": 0.038866378366947174,
      "epoch": 3.73,
      "learning_rate": 2.247369857743362e-05,
      "loss": 0.0449,
      "step": 3926,
      "task_loss": 0.0994512215256691
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7847768937426732,
      "compression_loss": 0.0,
      "distillation_loss": 0.14317655563354492,
      "epoch": 3.73,
      "learning_rate": 2.2463098104189497e-05,
      "loss": 0.1352,
      "step": 3927,
      "task_loss": 0.06382149457931519
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.78478704473562,
      "compression_loss": 0.0,
      "distillation_loss": 0.06667514145374298,
      "epoch": 3.73,
      "learning_rate": 2.2452498091785825e-05,
      "loss": 0.0739,
      "step": 3928,
      "task_loss": 0.13881635665893555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7847971912150083,
      "compression_loss": 0.0,
      "distillation_loss": 0.015922335907816887,
      "epoch": 3.73,
      "learning_rate": 2.244189854214814e-05,
      "loss": 0.0232,
      "step": 3929,
      "task_loss": 0.08892001956701279
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7848073331818419,
      "compression_loss": 0.0,
      "distillation_loss": 0.12270977348089218,
      "epoch": 3.73,
      "learning_rate": 2.24312994572019e-05,
      "loss": 0.1286,
      "step": 3930,
      "task_loss": 0.1818765252828598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7848174706371246,
      "compression_loss": 0.0,
      "distillation_loss": 0.09202725440263748,
      "epoch": 3.73,
      "learning_rate": 2.2420700838872493e-05,
      "loss": 0.0922,
      "step": 3931,
      "task_loss": 0.09422404319047928
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.78482760358186,
      "compression_loss": 0.0,
      "distillation_loss": 0.22878801822662354,
      "epoch": 3.73,
      "learning_rate": 2.2410102689085185e-05,
      "loss": 0.2324,
      "step": 3932,
      "task_loss": 0.2648504376411438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7848377320170516,
      "compression_loss": 0.0,
      "distillation_loss": 0.03465723991394043,
      "epoch": 3.74,
      "learning_rate": 2.2399505009765214e-05,
      "loss": 0.0383,
      "step": 3933,
      "task_loss": 0.07061752676963806
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7848478559437033,
      "compression_loss": 0.0,
      "distillation_loss": 0.046556901186704636,
      "epoch": 3.74,
      "learning_rate": 2.2388907802837676e-05,
      "loss": 0.0425,
      "step": 3934,
      "task_loss": 0.005850574001669884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7848579753628188,
      "compression_loss": 0.0,
      "distillation_loss": 0.02604757435619831,
      "epoch": 3.74,
      "learning_rate": 2.2378311070227616e-05,
      "loss": 0.0313,
      "step": 3935,
      "task_loss": 0.07850679755210876
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7848680902754016,
      "compression_loss": 0.0,
      "distillation_loss": 0.032080747187137604,
      "epoch": 3.74,
      "learning_rate": 2.2367714813859967e-05,
      "loss": 0.0366,
      "step": 3936,
      "task_loss": 0.07736522704362869
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7848782006824555,
      "compression_loss": 0.0,
      "distillation_loss": 0.030702613294124603,
      "epoch": 3.74,
      "learning_rate": 2.23571190356596e-05,
      "loss": 0.0332,
      "step": 3937,
      "task_loss": 0.05548100546002388
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7848883065849842,
      "compression_loss": 0.0,
      "distillation_loss": 0.06845727562904358,
      "epoch": 3.74,
      "learning_rate": 2.2346523737551296e-05,
      "loss": 0.0701,
      "step": 3938,
      "task_loss": 0.08478248119354248
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7848984079839914,
      "compression_loss": 0.0,
      "distillation_loss": 0.030870838090777397,
      "epoch": 3.74,
      "learning_rate": 2.233592892145973e-05,
      "loss": 0.0387,
      "step": 3939,
      "task_loss": 0.10916754603385925
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7849085048804807,
      "compression_loss": 0.0,
      "distillation_loss": 0.1040521040558815,
      "epoch": 3.74,
      "learning_rate": 2.23253345893095e-05,
      "loss": 0.1096,
      "step": 3940,
      "task_loss": 0.15946507453918457
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7849185972754558,
      "compression_loss": 0.0,
      "distillation_loss": 0.04144272953271866,
      "epoch": 3.74,
      "learning_rate": 2.231474074302513e-05,
      "loss": 0.051,
      "step": 3941,
      "task_loss": 0.13734087347984314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7849286851699203,
      "compression_loss": 0.0,
      "distillation_loss": 0.11080522835254669,
      "epoch": 3.74,
      "learning_rate": 2.2304147384531038e-05,
      "loss": 0.115,
      "step": 3942,
      "task_loss": 0.15283827483654022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7849387685648781,
      "compression_loss": 0.0,
      "distillation_loss": 0.07673658430576324,
      "epoch": 3.74,
      "learning_rate": 2.2293554515751552e-05,
      "loss": 0.08,
      "step": 3943,
      "task_loss": 0.10954262316226959
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7849488474613326,
      "compression_loss": 0.0,
      "distillation_loss": 0.13339099287986755,
      "epoch": 3.75,
      "learning_rate": 2.2282962138610924e-05,
      "loss": 0.1443,
      "step": 3944,
      "task_loss": 0.2422405332326889
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7849589218602878,
      "compression_loss": 0.0,
      "distillation_loss": 0.05010555312037468,
      "epoch": 3.75,
      "learning_rate": 2.2272370255033314e-05,
      "loss": 0.058,
      "step": 3945,
      "task_loss": 0.1292436569929123
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7849689917627473,
      "compression_loss": 0.0,
      "distillation_loss": 0.04166126996278763,
      "epoch": 3.75,
      "learning_rate": 2.226177886694278e-05,
      "loss": 0.0422,
      "step": 3946,
      "task_loss": 0.04674288257956505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7849790571697145,
      "compression_loss": 0.0,
      "distillation_loss": 0.05979571491479874,
      "epoch": 3.75,
      "learning_rate": 2.22511879762633e-05,
      "loss": 0.0623,
      "step": 3947,
      "task_loss": 0.08474962413311005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7849891180821934,
      "compression_loss": 0.0,
      "distillation_loss": 0.09109986573457718,
      "epoch": 3.75,
      "learning_rate": 2.2240597584918768e-05,
      "loss": 0.0975,
      "step": 3948,
      "task_loss": 0.155453622341156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7849991745011875,
      "compression_loss": 0.0,
      "distillation_loss": 0.07806719839572906,
      "epoch": 3.75,
      "learning_rate": 2.2230007694832963e-05,
      "loss": 0.0722,
      "step": 3949,
      "task_loss": 0.018949225544929504
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7850092264277005,
      "compression_loss": 0.0,
      "distillation_loss": 0.03633013740181923,
      "epoch": 3.75,
      "learning_rate": 2.2219418307929607e-05,
      "loss": 0.0392,
      "step": 3950,
      "task_loss": 0.06507201492786407
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7850192738627363,
      "compression_loss": 0.0,
      "distillation_loss": 0.029751278460025787,
      "epoch": 3.75,
      "learning_rate": 2.2208829426132307e-05,
      "loss": 0.0389,
      "step": 3951,
      "task_loss": 0.12085320800542831
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7850293168072984,
      "compression_loss": 0.0,
      "distillation_loss": 0.056585509330034256,
      "epoch": 3.75,
      "learning_rate": 2.219824105136459e-05,
      "loss": 0.0607,
      "step": 3952,
      "task_loss": 0.0976976677775383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7850393552623904,
      "compression_loss": 0.0,
      "distillation_loss": 0.029133249074220657,
      "epoch": 3.75,
      "learning_rate": 2.218765318554987e-05,
      "loss": 0.0329,
      "step": 3953,
      "task_loss": 0.06695835292339325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7850493892290161,
      "compression_loss": 0.0,
      "distillation_loss": 0.04859749227762222,
      "epoch": 3.75,
      "learning_rate": 2.2177065830611498e-05,
      "loss": 0.046,
      "step": 3954,
      "task_loss": 0.022643616423010826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7850594187081792,
      "compression_loss": 0.0,
      "distillation_loss": 0.0373307503759861,
      "epoch": 3.76,
      "learning_rate": 2.2166478988472716e-05,
      "loss": 0.0352,
      "step": 3955,
      "task_loss": 0.015915043652057648
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7850694437008833,
      "compression_loss": 0.0,
      "distillation_loss": 0.05053357779979706,
      "epoch": 3.76,
      "learning_rate": 2.215589266105667e-05,
      "loss": 0.0505,
      "step": 3956,
      "task_loss": 0.049804605543613434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7850794642081321,
      "compression_loss": 0.0,
      "distillation_loss": 0.016781002283096313,
      "epoch": 3.76,
      "learning_rate": 2.2145306850286424e-05,
      "loss": 0.0155,
      "step": 3957,
      "task_loss": 0.00426999107003212
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7850894802309294,
      "compression_loss": 0.0,
      "distillation_loss": 0.06677426397800446,
      "epoch": 3.76,
      "learning_rate": 2.2134721558084917e-05,
      "loss": 0.0677,
      "step": 3958,
      "task_loss": 0.07622140645980835
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7850994917702788,
      "compression_loss": 0.0,
      "distillation_loss": 0.031092319637537003,
      "epoch": 3.76,
      "learning_rate": 2.2124136786375057e-05,
      "loss": 0.0348,
      "step": 3959,
      "task_loss": 0.06810992956161499
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.785109498827184,
      "compression_loss": 0.0,
      "distillation_loss": 0.04789033159613609,
      "epoch": 3.76,
      "learning_rate": 2.2113552537079597e-05,
      "loss": 0.0456,
      "step": 3960,
      "task_loss": 0.025089185684919357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7851195014026485,
      "compression_loss": 0.0,
      "distillation_loss": 0.03125907480716705,
      "epoch": 3.76,
      "learning_rate": 2.2102968812121218e-05,
      "loss": 0.0292,
      "step": 3961,
      "task_loss": 0.011070974171161652
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7851294994976762,
      "compression_loss": 0.0,
      "distillation_loss": 0.0916096568107605,
      "epoch": 3.76,
      "learning_rate": 2.209238561342251e-05,
      "loss": 0.0876,
      "step": 3962,
      "task_loss": 0.05148671194911003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7851394931132707,
      "compression_loss": 0.0,
      "distillation_loss": 0.17331336438655853,
      "epoch": 3.76,
      "learning_rate": 2.2081802942905955e-05,
      "loss": 0.1803,
      "step": 3963,
      "task_loss": 0.24269163608551025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7851494822504358,
      "compression_loss": 0.0,
      "distillation_loss": 0.022669488564133644,
      "epoch": 3.76,
      "learning_rate": 2.2071220802493954e-05,
      "loss": 0.0208,
      "step": 3964,
      "task_loss": 0.0038123298436403275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.785159466910175,
      "compression_loss": 0.0,
      "distillation_loss": 0.07016949355602264,
      "epoch": 3.77,
      "learning_rate": 2.2060639194108794e-05,
      "loss": 0.0671,
      "step": 3965,
      "task_loss": 0.039382204413414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.785169447093492,
      "compression_loss": 0.0,
      "distillation_loss": 0.05281013250350952,
      "epoch": 3.77,
      "learning_rate": 2.2050058119672677e-05,
      "loss": 0.0531,
      "step": 3966,
      "task_loss": 0.05565192550420761
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7851794228013906,
      "compression_loss": 0.0,
      "distillation_loss": 0.1318967640399933,
      "epoch": 3.77,
      "learning_rate": 2.2039477581107714e-05,
      "loss": 0.1256,
      "step": 3967,
      "task_loss": 0.06887546926736832
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7851893940348745,
      "compression_loss": 0.0,
      "distillation_loss": 0.0660499557852745,
      "epoch": 3.77,
      "learning_rate": 2.2028897580335906e-05,
      "loss": 0.0626,
      "step": 3968,
      "task_loss": 0.03172999620437622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7851993607949473,
      "compression_loss": 0.0,
      "distillation_loss": 0.03410178795456886,
      "epoch": 3.77,
      "learning_rate": 2.2018318119279168e-05,
      "loss": 0.0312,
      "step": 3969,
      "task_loss": 0.004658637568354607
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7852093230826127,
      "compression_loss": 0.0,
      "distillation_loss": 0.16602616012096405,
      "epoch": 3.77,
      "learning_rate": 2.2007739199859294e-05,
      "loss": 0.169,
      "step": 3970,
      "task_loss": 0.19587615132331848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7852192808988743,
      "compression_loss": 0.0,
      "distillation_loss": 0.05224110186100006,
      "epoch": 3.77,
      "learning_rate": 2.1997160823998016e-05,
      "loss": 0.0481,
      "step": 3971,
      "task_loss": 0.01118180900812149
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7852292342447359,
      "compression_loss": 0.0,
      "distillation_loss": 0.055276624858379364,
      "epoch": 3.77,
      "learning_rate": 2.1986582993616926e-05,
      "loss": 0.0502,
      "step": 3972,
      "task_loss": 0.004159906879067421
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7852391831212011,
      "compression_loss": 0.0,
      "distillation_loss": 0.05159646272659302,
      "epoch": 3.77,
      "learning_rate": 2.1976005710637554e-05,
      "loss": 0.0558,
      "step": 3973,
      "task_loss": 0.09320049732923508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7852491275292737,
      "compression_loss": 0.0,
      "distillation_loss": 0.10250347852706909,
      "epoch": 3.77,
      "learning_rate": 2.196542897698131e-05,
      "loss": 0.1108,
      "step": 3974,
      "task_loss": 0.185111865401268
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7852590674699573,
      "compression_loss": 0.0,
      "distillation_loss": 0.04058488458395004,
      "epoch": 3.77,
      "learning_rate": 2.1954852794569493e-05,
      "loss": 0.0404,
      "step": 3975,
      "task_loss": 0.03915491700172424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7852690029442555,
      "compression_loss": 0.0,
      "distillation_loss": 0.012665364891290665,
      "epoch": 3.78,
      "learning_rate": 2.194427716532334e-05,
      "loss": 0.0117,
      "step": 3976,
      "task_loss": 0.0031289253383874893
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7852789339531722,
      "compression_loss": 0.0,
      "distillation_loss": 0.07077420502901077,
      "epoch": 3.78,
      "learning_rate": 2.193370209116396e-05,
      "loss": 0.0676,
      "step": 3977,
      "task_loss": 0.038839004933834076
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7852888604977108,
      "compression_loss": 0.0,
      "distillation_loss": 0.039995912462472916,
      "epoch": 3.78,
      "learning_rate": 2.1923127574012365e-05,
      "loss": 0.0426,
      "step": 3978,
      "task_loss": 0.06648865342140198
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7852987825788752,
      "compression_loss": 0.0,
      "distillation_loss": 0.06262241303920746,
      "epoch": 3.78,
      "learning_rate": 2.1912553615789462e-05,
      "loss": 0.0587,
      "step": 3979,
      "task_loss": 0.023578649386763573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7853087001976691,
      "compression_loss": 0.0,
      "distillation_loss": 0.05083940178155899,
      "epoch": 3.78,
      "learning_rate": 2.190198021841606e-05,
      "loss": 0.0518,
      "step": 3980,
      "task_loss": 0.05996212735772133
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7853186133550959,
      "compression_loss": 0.0,
      "distillation_loss": 0.0958191305398941,
      "epoch": 3.78,
      "learning_rate": 2.189140738381288e-05,
      "loss": 0.0944,
      "step": 3981,
      "task_loss": 0.08185603469610214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7853285220521596,
      "compression_loss": 0.0,
      "distillation_loss": 0.1051722839474678,
      "epoch": 3.78,
      "learning_rate": 2.188083511390051e-05,
      "loss": 0.1112,
      "step": 3982,
      "task_loss": 0.1649647355079651
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7853384262898637,
      "compression_loss": 0.0,
      "distillation_loss": 0.10239297151565552,
      "epoch": 3.78,
      "learning_rate": 2.1870263410599464e-05,
      "loss": 0.1246,
      "step": 3983,
      "task_loss": 0.3248556852340698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.785348326069212,
      "compression_loss": 0.0,
      "distillation_loss": 0.04012586921453476,
      "epoch": 3.78,
      "learning_rate": 2.185969227583014e-05,
      "loss": 0.0377,
      "step": 3984,
      "task_loss": 0.015774773433804512
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7853582213912081,
      "compression_loss": 0.0,
      "distillation_loss": 0.019525595009326935,
      "epoch": 3.78,
      "learning_rate": 2.1849121711512847e-05,
      "loss": 0.0185,
      "step": 3985,
      "task_loss": 0.009618887677788734
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7853681122568558,
      "compression_loss": 0.0,
      "distillation_loss": 0.11984270811080933,
      "epoch": 3.79,
      "learning_rate": 2.1838551719567767e-05,
      "loss": 0.1232,
      "step": 3986,
      "task_loss": 0.1538366973400116
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7853779986671586,
      "compression_loss": 0.0,
      "distillation_loss": 0.10699717700481415,
      "epoch": 3.79,
      "learning_rate": 2.1827982301914993e-05,
      "loss": 0.1009,
      "step": 3987,
      "task_loss": 0.04600843787193298
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7853878806231203,
      "compression_loss": 0.0,
      "distillation_loss": 0.018371235579252243,
      "epoch": 3.79,
      "learning_rate": 2.1817413460474514e-05,
      "loss": 0.0415,
      "step": 3988,
      "task_loss": 0.2499411404132843
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7853977581257445,
      "compression_loss": 0.0,
      "distillation_loss": 0.07233263552188873,
      "epoch": 3.79,
      "learning_rate": 2.18068451971662e-05,
      "loss": 0.0799,
      "step": 3989,
      "task_loss": 0.14828746020793915
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.785407631176035,
      "compression_loss": 0.0,
      "distillation_loss": 0.04718535393476486,
      "epoch": 3.79,
      "learning_rate": 2.1796277513909838e-05,
      "loss": 0.0454,
      "step": 3990,
      "task_loss": 0.029316851869225502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7854174997749954,
      "compression_loss": 0.0,
      "distillation_loss": 0.03916257619857788,
      "epoch": 3.79,
      "learning_rate": 2.1785710412625098e-05,
      "loss": 0.0363,
      "step": 3991,
      "task_loss": 0.010277681052684784
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7854273639236293,
      "compression_loss": 0.0,
      "distillation_loss": 0.179625004529953,
      "epoch": 3.79,
      "learning_rate": 2.1775143895231533e-05,
      "loss": 0.1678,
      "step": 3992,
      "task_loss": 0.0614137165248394
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7854372236229407,
      "compression_loss": 0.0,
      "distillation_loss": 0.11892496049404144,
      "epoch": 3.79,
      "learning_rate": 2.1764577963648614e-05,
      "loss": 0.1157,
      "step": 3993,
      "task_loss": 0.08710844814777374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7854470788739328,
      "compression_loss": 0.0,
      "distillation_loss": 0.04435303807258606,
      "epoch": 3.79,
      "learning_rate": 2.175401261979569e-05,
      "loss": 0.0626,
      "step": 3994,
      "task_loss": 0.2271975576877594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7854569296776097,
      "compression_loss": 0.0,
      "distillation_loss": 0.04650871828198433,
      "epoch": 3.79,
      "learning_rate": 2.1743447865592016e-05,
      "loss": 0.0441,
      "step": 3995,
      "task_loss": 0.022587845101952553
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.785466776034975,
      "compression_loss": 0.0,
      "distillation_loss": 0.09561876952648163,
      "epoch": 3.79,
      "learning_rate": 2.173288370295671e-05,
      "loss": 0.0919,
      "step": 3996,
      "task_loss": 0.058294668793678284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7854766179470322,
      "compression_loss": 0.0,
      "distillation_loss": 0.054196953773498535,
      "epoch": 3.8,
      "learning_rate": 2.1722320133808815e-05,
      "loss": 0.0555,
      "step": 3997,
      "task_loss": 0.06700082123279572
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7854864554147851,
      "compression_loss": 0.0,
      "distillation_loss": 0.09690621495246887,
      "epoch": 3.8,
      "learning_rate": 2.171175716006726e-05,
      "loss": 0.1012,
      "step": 3998,
      "task_loss": 0.14022237062454224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7854962884392374,
      "compression_loss": 0.0,
      "distillation_loss": 0.04198819026350975,
      "epoch": 3.8,
      "learning_rate": 2.1701194783650846e-05,
      "loss": 0.0455,
      "step": 3999,
      "task_loss": 0.07720094919204712
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7855061170213927,
      "compression_loss": 0.0,
      "distillation_loss": 0.12908399105072021,
      "epoch": 3.8,
      "learning_rate": 2.1690633006478293e-05,
      "loss": 0.1321,
      "step": 4000,
      "task_loss": 0.15927539765834808
    },
    {
      "epoch": 3.8,
      "eval_accuracy": 0.8944954128440367,
      "eval_loss": 0.41921311616897583,
      "eval_runtime": 18.4061,
      "eval_samples_per_second": 47.376,
      "eval_steps_per_second": 5.922,
      "step": 4000
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7855159411622548,
      "compression_loss": 0.0,
      "distillation_loss": 0.0658789873123169,
      "epoch": 3.8,
      "learning_rate": 2.1680071830468178e-05,
      "loss": 0.0634,
      "step": 4001,
      "task_loss": 0.04087566211819649
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7855257608628273,
      "compression_loss": 0.0,
      "distillation_loss": 0.06824058294296265,
      "epoch": 3.8,
      "learning_rate": 2.166951125753902e-05,
      "loss": 0.0721,
      "step": 4002,
      "task_loss": 0.10674886405467987
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7855355761241138,
      "compression_loss": 0.0,
      "distillation_loss": 0.12798132002353668,
      "epoch": 3.8,
      "learning_rate": 2.1658951289609174e-05,
      "loss": 0.1246,
      "step": 4003,
      "task_loss": 0.09367252886295319
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7855453869471183,
      "compression_loss": 0.0,
      "distillation_loss": 0.15844711661338806,
      "epoch": 3.8,
      "learning_rate": 2.1648391928596917e-05,
      "loss": 0.1684,
      "step": 4004,
      "task_loss": 0.25770193338394165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7855551933328442,
      "compression_loss": 0.0,
      "distillation_loss": 0.029653755947947502,
      "epoch": 3.8,
      "learning_rate": 2.1637833176420417e-05,
      "loss": 0.0273,
      "step": 4005,
      "task_loss": 0.005788305774331093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7855649952822952,
      "compression_loss": 0.0,
      "distillation_loss": 0.07723814994096756,
      "epoch": 3.8,
      "learning_rate": 2.1627275034997704e-05,
      "loss": 0.074,
      "step": 4006,
      "task_loss": 0.04511658847332001
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.785574792796475,
      "compression_loss": 0.0,
      "distillation_loss": 0.03552757203578949,
      "epoch": 3.81,
      "learning_rate": 2.161671750624673e-05,
      "loss": 0.0364,
      "step": 4007,
      "task_loss": 0.04422738030552864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7855845858763874,
      "compression_loss": 0.0,
      "distillation_loss": 0.02493412420153618,
      "epoch": 3.81,
      "learning_rate": 2.160616059208531e-05,
      "loss": 0.0286,
      "step": 4008,
      "task_loss": 0.06140220910310745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7855943745230358,
      "compression_loss": 0.0,
      "distillation_loss": 0.024118445813655853,
      "epoch": 3.81,
      "learning_rate": 2.1595604294431158e-05,
      "loss": 0.041,
      "step": 4009,
      "task_loss": 0.19306635856628418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7856041587374243,
      "compression_loss": 0.0,
      "distillation_loss": 0.06833194196224213,
      "epoch": 3.81,
      "learning_rate": 2.1585048615201885e-05,
      "loss": 0.0668,
      "step": 4010,
      "task_loss": 0.05287961661815643
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7856139385205563,
      "compression_loss": 0.0,
      "distillation_loss": 0.08782090991735458,
      "epoch": 3.81,
      "learning_rate": 2.1574493556314983e-05,
      "loss": 0.1022,
      "step": 4011,
      "task_loss": 0.23139870166778564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7856237138734354,
      "compression_loss": 0.0,
      "distillation_loss": 0.03765169531106949,
      "epoch": 3.81,
      "learning_rate": 2.1563939119687828e-05,
      "loss": 0.0461,
      "step": 4012,
      "task_loss": 0.12255912274122238
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7856334847970656,
      "compression_loss": 0.0,
      "distillation_loss": 0.047839343547821045,
      "epoch": 3.81,
      "learning_rate": 2.155338530723767e-05,
      "loss": 0.0597,
      "step": 4013,
      "task_loss": 0.16615822911262512
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7856432512924504,
      "compression_loss": 0.0,
      "distillation_loss": 0.0834595113992691,
      "epoch": 3.81,
      "learning_rate": 2.154283212088168e-05,
      "loss": 0.0868,
      "step": 4014,
      "task_loss": 0.11699174344539642
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7856530133605933,
      "compression_loss": 0.0,
      "distillation_loss": 0.028070103377103806,
      "epoch": 3.81,
      "learning_rate": 2.1532279562536874e-05,
      "loss": 0.0308,
      "step": 4015,
      "task_loss": 0.05518289655447006
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7856627710024984,
      "compression_loss": 0.0,
      "distillation_loss": 0.019197819754481316,
      "epoch": 3.81,
      "learning_rate": 2.1521727634120192e-05,
      "loss": 0.0184,
      "step": 4016,
      "task_loss": 0.011533919721841812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.785672524219169,
      "compression_loss": 0.0,
      "distillation_loss": 0.028306379914283752,
      "epoch": 3.81,
      "learning_rate": 2.151117633754844e-05,
      "loss": 0.0421,
      "step": 4017,
      "task_loss": 0.1661168932914734
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7856822730116091,
      "compression_loss": 0.0,
      "distillation_loss": 0.026535285636782646,
      "epoch": 3.82,
      "learning_rate": 2.150062567473829e-05,
      "loss": 0.0298,
      "step": 4018,
      "task_loss": 0.058951202780008316
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7856920173808221,
      "compression_loss": 0.0,
      "distillation_loss": 0.040127944201231,
      "epoch": 3.82,
      "learning_rate": 2.1490075647606363e-05,
      "loss": 0.0386,
      "step": 4019,
      "task_loss": 0.025223994627594948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7857017573278118,
      "compression_loss": 0.0,
      "distillation_loss": 0.02355261519551277,
      "epoch": 3.82,
      "learning_rate": 2.1479526258069087e-05,
      "loss": 0.0352,
      "step": 4020,
      "task_loss": 0.13954126834869385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7857114928535818,
      "compression_loss": 0.0,
      "distillation_loss": 0.03306501358747482,
      "epoch": 3.82,
      "learning_rate": 2.1468977508042824e-05,
      "loss": 0.0479,
      "step": 4021,
      "task_loss": 0.1817673295736313
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.785721223959136,
      "compression_loss": 0.0,
      "distillation_loss": 0.06848086416721344,
      "epoch": 3.82,
      "learning_rate": 2.14584293994438e-05,
      "loss": 0.0723,
      "step": 4022,
      "task_loss": 0.10678940266370773
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7857309506454778,
      "compression_loss": 0.0,
      "distillation_loss": 0.03855840116739273,
      "epoch": 3.82,
      "learning_rate": 2.1447881934188134e-05,
      "loss": 0.0358,
      "step": 4023,
      "task_loss": 0.010923238471150398
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7857406729136112,
      "compression_loss": 0.0,
      "distillation_loss": 0.021052051335573196,
      "epoch": 3.82,
      "learning_rate": 2.143733511419183e-05,
      "loss": 0.0193,
      "step": 4024,
      "task_loss": 0.003061491996049881
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7857503907645396,
      "compression_loss": 0.0,
      "distillation_loss": 0.10116279125213623,
      "epoch": 3.82,
      "learning_rate": 2.142678894137075e-05,
      "loss": 0.1047,
      "step": 4025,
      "task_loss": 0.1363639235496521
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7857601041992668,
      "compression_loss": 0.0,
      "distillation_loss": 0.19848263263702393,
      "epoch": 3.82,
      "learning_rate": 2.1416243417640668e-05,
      "loss": 0.1934,
      "step": 4026,
      "task_loss": 0.14759844541549683
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7857698132187965,
      "compression_loss": 0.0,
      "distillation_loss": 0.10897897183895111,
      "epoch": 3.82,
      "learning_rate": 2.1405698544917225e-05,
      "loss": 0.1032,
      "step": 4027,
      "task_loss": 0.051134128123521805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7857795178241324,
      "compression_loss": 0.0,
      "distillation_loss": 0.06620095670223236,
      "epoch": 3.83,
      "learning_rate": 2.139515432511596e-05,
      "loss": 0.0624,
      "step": 4028,
      "task_loss": 0.0285488348454237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.785789218016278,
      "compression_loss": 0.0,
      "distillation_loss": 0.1283935308456421,
      "epoch": 3.83,
      "learning_rate": 2.138461076015227e-05,
      "loss": 0.1225,
      "step": 4029,
      "task_loss": 0.0697251558303833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7857989137962372,
      "compression_loss": 0.0,
      "distillation_loss": 0.05554177612066269,
      "epoch": 3.83,
      "learning_rate": 2.1374067851941445e-05,
      "loss": 0.0517,
      "step": 4030,
      "task_loss": 0.017079105600714684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7858086051650136,
      "compression_loss": 0.0,
      "distillation_loss": 0.015999089926481247,
      "epoch": 3.83,
      "learning_rate": 2.1363525602398666e-05,
      "loss": 0.0233,
      "step": 4031,
      "task_loss": 0.08854260295629501
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.785818292123611,
      "compression_loss": 0.0,
      "distillation_loss": 0.12333665788173676,
      "epoch": 3.83,
      "learning_rate": 2.1352984013438964e-05,
      "loss": 0.1226,
      "step": 4032,
      "task_loss": 0.11570969223976135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7858279746730328,
      "compression_loss": 0.0,
      "distillation_loss": 0.015158230438828468,
      "epoch": 3.83,
      "learning_rate": 2.134244308697729e-05,
      "loss": 0.0222,
      "step": 4033,
      "task_loss": 0.08555779606103897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7858376528142829,
      "compression_loss": 0.0,
      "distillation_loss": 0.06008541211485863,
      "epoch": 3.83,
      "learning_rate": 2.133190282492844e-05,
      "loss": 0.064,
      "step": 4034,
      "task_loss": 0.09927615523338318
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7858473265483649,
      "compression_loss": 0.0,
      "distillation_loss": 0.1671043038368225,
      "epoch": 3.83,
      "learning_rate": 2.1321363229207096e-05,
      "loss": 0.1689,
      "step": 4035,
      "task_loss": 0.1853921115398407
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7858569958762825,
      "compression_loss": 0.0,
      "distillation_loss": 0.026313871145248413,
      "epoch": 3.83,
      "learning_rate": 2.131082430172785e-05,
      "loss": 0.0243,
      "step": 4036,
      "task_loss": 0.006411956623196602
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7858666607990394,
      "compression_loss": 0.0,
      "distillation_loss": 0.07281780242919922,
      "epoch": 3.83,
      "learning_rate": 2.1300286044405135e-05,
      "loss": 0.0689,
      "step": 4037,
      "task_loss": 0.033516231924295425
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7858763213176394,
      "compression_loss": 0.0,
      "distillation_loss": 0.04494217038154602,
      "epoch": 3.83,
      "learning_rate": 2.1289748459153283e-05,
      "loss": 0.0421,
      "step": 4038,
      "task_loss": 0.016057439148426056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.785885977433086,
      "compression_loss": 0.0,
      "distillation_loss": 0.02288595214486122,
      "epoch": 3.84,
      "learning_rate": 2.1279211547886485e-05,
      "loss": 0.0211,
      "step": 4039,
      "task_loss": 0.005246007815003395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7858956291463829,
      "compression_loss": 0.0,
      "distillation_loss": 0.041702110320329666,
      "epoch": 3.84,
      "learning_rate": 2.1268675312518833e-05,
      "loss": 0.0515,
      "step": 4040,
      "task_loss": 0.1393231451511383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7859052764585338,
      "compression_loss": 0.0,
      "distillation_loss": 0.034420765936374664,
      "epoch": 3.84,
      "learning_rate": 2.1258139754964283e-05,
      "loss": 0.0314,
      "step": 4041,
      "task_loss": 0.004395313560962677
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7859149193705425,
      "compression_loss": 0.0,
      "distillation_loss": 0.057688407599925995,
      "epoch": 3.84,
      "learning_rate": 2.124760487713666e-05,
      "loss": 0.0542,
      "step": 4042,
      "task_loss": 0.022908154875040054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7859245578834124,
      "compression_loss": 0.0,
      "distillation_loss": 0.06689159572124481,
      "epoch": 3.84,
      "learning_rate": 2.1237070680949686e-05,
      "loss": 0.0685,
      "step": 4043,
      "task_loss": 0.08265194296836853
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7859341919981475,
      "compression_loss": 0.0,
      "distillation_loss": 0.03234408050775528,
      "epoch": 3.84,
      "learning_rate": 2.122653716831694e-05,
      "loss": 0.0304,
      "step": 4044,
      "task_loss": 0.012456223368644714
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7859438217157514,
      "compression_loss": 0.0,
      "distillation_loss": 0.046421635895967484,
      "epoch": 3.84,
      "learning_rate": 2.12160043411519e-05,
      "loss": 0.0521,
      "step": 4045,
      "task_loss": 0.10356031358242035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7859534470372276,
      "compression_loss": 0.0,
      "distillation_loss": 0.26599907875061035,
      "epoch": 3.84,
      "learning_rate": 2.120547220136789e-05,
      "loss": 0.2447,
      "step": 4046,
      "task_loss": 0.053442008793354034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.78596306796358,
      "compression_loss": 0.0,
      "distillation_loss": 0.06889566034078598,
      "epoch": 3.84,
      "learning_rate": 2.1194940750878132e-05,
      "loss": 0.0723,
      "step": 4047,
      "task_loss": 0.10272115468978882
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7859726844958121,
      "compression_loss": 0.0,
      "distillation_loss": 0.03883613273501396,
      "epoch": 3.84,
      "learning_rate": 2.1184409991595713e-05,
      "loss": 0.0432,
      "step": 4048,
      "task_loss": 0.0823683887720108
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7859822966349278,
      "compression_loss": 0.0,
      "distillation_loss": 0.0397968553006649,
      "epoch": 3.85,
      "learning_rate": 2.117387992543359e-05,
      "loss": 0.0434,
      "step": 4049,
      "task_loss": 0.07577399909496307
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7859919043819307,
      "compression_loss": 0.0,
      "distillation_loss": 0.17865076661109924,
      "epoch": 3.85,
      "learning_rate": 2.1163350554304613e-05,
      "loss": 0.1735,
      "step": 4050,
      "task_loss": 0.12746202945709229
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7860015077378243,
      "compression_loss": 0.0,
      "distillation_loss": 0.02501235529780388,
      "epoch": 3.85,
      "learning_rate": 2.1152821880121482e-05,
      "loss": 0.0232,
      "step": 4051,
      "task_loss": 0.007309248670935631
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7860111067036125,
      "compression_loss": 0.0,
      "distillation_loss": 0.07480224221944809,
      "epoch": 3.85,
      "learning_rate": 2.1142293904796783e-05,
      "loss": 0.0842,
      "step": 4052,
      "task_loss": 0.16905774176120758
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7860207012802989,
      "compression_loss": 0.0,
      "distillation_loss": 0.04464121162891388,
      "epoch": 3.85,
      "learning_rate": 2.1131766630242966e-05,
      "loss": 0.0478,
      "step": 4053,
      "task_loss": 0.07629342377185822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7860302914688873,
      "compression_loss": 0.0,
      "distillation_loss": 0.023911893367767334,
      "epoch": 3.85,
      "learning_rate": 2.112124005837238e-05,
      "loss": 0.0219,
      "step": 4054,
      "task_loss": 0.004040185362100601
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.786039877270381,
      "compression_loss": 0.0,
      "distillation_loss": 0.017198091372847557,
      "epoch": 3.85,
      "learning_rate": 2.1110714191097222e-05,
      "loss": 0.0159,
      "step": 4055,
      "task_loss": 0.004315854981541634
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7860494586857841,
      "compression_loss": 0.0,
      "distillation_loss": 0.09507577121257782,
      "epoch": 3.85,
      "learning_rate": 2.1100189030329558e-05,
      "loss": 0.0988,
      "step": 4056,
      "task_loss": 0.1326824277639389
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7860590357161003,
      "compression_loss": 0.0,
      "distillation_loss": 0.03281257301568985,
      "epoch": 3.85,
      "learning_rate": 2.108966457798134e-05,
      "loss": 0.0358,
      "step": 4057,
      "task_loss": 0.06237972527742386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7860686083623328,
      "compression_loss": 0.0,
      "distillation_loss": 0.09298193454742432,
      "epoch": 3.85,
      "learning_rate": 2.107914083596438e-05,
      "loss": 0.1066,
      "step": 4058,
      "task_loss": 0.22964733839035034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7860781766254858,
      "compression_loss": 0.0,
      "distillation_loss": 0.10477671027183533,
      "epoch": 3.85,
      "learning_rate": 2.106861780619037e-05,
      "loss": 0.0998,
      "step": 4059,
      "task_loss": 0.054976824671030045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7860877405065628,
      "compression_loss": 0.0,
      "distillation_loss": 0.025005143135786057,
      "epoch": 3.86,
      "learning_rate": 2.105809549057088e-05,
      "loss": 0.0288,
      "step": 4060,
      "task_loss": 0.06275138258934021
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7860973000065674,
      "compression_loss": 0.0,
      "distillation_loss": 0.06488427519798279,
      "epoch": 3.86,
      "learning_rate": 2.1047573891017306e-05,
      "loss": 0.0644,
      "step": 4061,
      "task_loss": 0.06053687259554863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7861068551265034,
      "compression_loss": 0.0,
      "distillation_loss": 0.13458393514156342,
      "epoch": 3.86,
      "learning_rate": 2.103705300944099e-05,
      "loss": 0.1369,
      "step": 4062,
      "task_loss": 0.15822012722492218
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7861164058673743,
      "compression_loss": 0.0,
      "distillation_loss": 0.18830344080924988,
      "epoch": 3.86,
      "learning_rate": 2.1026532847753068e-05,
      "loss": 0.181,
      "step": 4063,
      "task_loss": 0.11519521474838257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.786125952230184,
      "compression_loss": 0.0,
      "distillation_loss": 0.044180721044540405,
      "epoch": 3.86,
      "learning_rate": 2.1016013407864605e-05,
      "loss": 0.0402,
      "step": 4064,
      "task_loss": 0.004225835204124451
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7861354942159361,
      "compression_loss": 0.0,
      "distillation_loss": 0.03563191741704941,
      "epoch": 3.86,
      "learning_rate": 2.1005494691686482e-05,
      "loss": 0.0328,
      "step": 4065,
      "task_loss": 0.007603077217936516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7861450318256342,
      "compression_loss": 0.0,
      "distillation_loss": 0.024195533245801926,
      "epoch": 3.86,
      "learning_rate": 2.0994976701129488e-05,
      "loss": 0.0302,
      "step": 4066,
      "task_loss": 0.08473625034093857
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7861545650602821,
      "compression_loss": 0.0,
      "distillation_loss": 0.02827250212430954,
      "epoch": 3.86,
      "learning_rate": 2.098445943810427e-05,
      "loss": 0.026,
      "step": 4067,
      "task_loss": 0.005987679585814476
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7861640939208835,
      "compression_loss": 0.0,
      "distillation_loss": 0.026403294876217842,
      "epoch": 3.86,
      "learning_rate": 2.0973942904521328e-05,
      "loss": 0.0253,
      "step": 4068,
      "task_loss": 0.0155414380133152
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.786173618408442,
      "compression_loss": 0.0,
      "distillation_loss": 0.11837053298950195,
      "epoch": 3.86,
      "learning_rate": 2.096342710229105e-05,
      "loss": 0.1148,
      "step": 4069,
      "task_loss": 0.08244297653436661
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7861831385239613,
      "compression_loss": 0.0,
      "distillation_loss": 0.04006524384021759,
      "epoch": 3.87,
      "learning_rate": 2.0952912033323672e-05,
      "loss": 0.0377,
      "step": 4070,
      "task_loss": 0.016373053193092346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7861926542684451,
      "compression_loss": 0.0,
      "distillation_loss": 0.0364631786942482,
      "epoch": 3.87,
      "learning_rate": 2.0942397699529325e-05,
      "loss": 0.0337,
      "step": 4071,
      "task_loss": 0.008914249017834663
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.786202165642897,
      "compression_loss": 0.0,
      "distillation_loss": 0.1414714902639389,
      "epoch": 3.87,
      "learning_rate": 2.0931884102817973e-05,
      "loss": 0.1359,
      "step": 4072,
      "task_loss": 0.08563140779733658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7862116726483208,
      "compression_loss": 0.0,
      "distillation_loss": 0.0371423065662384,
      "epoch": 3.87,
      "learning_rate": 2.0921371245099463e-05,
      "loss": 0.0401,
      "step": 4073,
      "task_loss": 0.06720311939716339
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7862211752857201,
      "compression_loss": 0.0,
      "distillation_loss": 0.2344522476196289,
      "epoch": 3.87,
      "learning_rate": 2.0910859128283517e-05,
      "loss": 0.2293,
      "step": 4074,
      "task_loss": 0.18332399427890778
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7862306735560987,
      "compression_loss": 0.0,
      "distillation_loss": 0.042274147272109985,
      "epoch": 3.87,
      "learning_rate": 2.0900347754279698e-05,
      "loss": 0.0531,
      "step": 4075,
      "task_loss": 0.15049678087234497
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.78624016746046,
      "compression_loss": 0.0,
      "distillation_loss": 0.045464303344488144,
      "epoch": 3.87,
      "learning_rate": 2.088983712499745e-05,
      "loss": 0.043,
      "step": 4076,
      "task_loss": 0.02061353251338005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.786249656999808,
      "compression_loss": 0.0,
      "distillation_loss": 0.13638713955879211,
      "epoch": 3.87,
      "learning_rate": 2.0879327242346093e-05,
      "loss": 0.132,
      "step": 4077,
      "task_loss": 0.09256385266780853
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7862591421751463,
      "compression_loss": 0.0,
      "distillation_loss": 0.07257235050201416,
      "epoch": 3.87,
      "learning_rate": 2.0868818108234783e-05,
      "loss": 0.078,
      "step": 4078,
      "task_loss": 0.12710356712341309
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7862686229874785,
      "compression_loss": 0.0,
      "distillation_loss": 0.1785210520029068,
      "epoch": 3.87,
      "learning_rate": 2.0858309724572554e-05,
      "loss": 0.1706,
      "step": 4079,
      "task_loss": 0.09951774030923843
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7862780994378082,
      "compression_loss": 0.0,
      "distillation_loss": 0.04436881095170975,
      "epoch": 3.87,
      "learning_rate": 2.084780209326831e-05,
      "loss": 0.0416,
      "step": 4080,
      "task_loss": 0.016401471570134163
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7862875715271395,
      "compression_loss": 0.0,
      "distillation_loss": 0.026207074522972107,
      "epoch": 3.88,
      "learning_rate": 2.0837295216230826e-05,
      "loss": 0.0242,
      "step": 4081,
      "task_loss": 0.006385818123817444
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7862970392564755,
      "compression_loss": 0.0,
      "distillation_loss": 0.07042695581912994,
      "epoch": 3.88,
      "learning_rate": 2.0826789095368705e-05,
      "loss": 0.0771,
      "step": 4082,
      "task_loss": 0.1371847689151764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7863065026268203,
      "compression_loss": 0.0,
      "distillation_loss": 0.09259558469057083,
      "epoch": 3.88,
      "learning_rate": 2.081628373259044e-05,
      "loss": 0.097,
      "step": 4083,
      "task_loss": 0.13651143014431
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7863159616391775,
      "compression_loss": 0.0,
      "distillation_loss": 0.1024775579571724,
      "epoch": 3.88,
      "learning_rate": 2.0805779129804397e-05,
      "loss": 0.0936,
      "step": 4084,
      "task_loss": 0.01337103545665741
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7863254162945507,
      "compression_loss": 0.0,
      "distillation_loss": 0.0833493024110794,
      "epoch": 3.88,
      "learning_rate": 2.0795275288918763e-05,
      "loss": 0.0871,
      "step": 4085,
      "task_loss": 0.12135336548089981
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7863348665939435,
      "compression_loss": 0.0,
      "distillation_loss": 0.03871447220444679,
      "epoch": 3.88,
      "learning_rate": 2.0784772211841624e-05,
      "loss": 0.0447,
      "step": 4086,
      "task_loss": 0.09859400987625122
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7863443125383598,
      "compression_loss": 0.0,
      "distillation_loss": 0.03221891075372696,
      "epoch": 3.88,
      "learning_rate": 2.077426990048091e-05,
      "loss": 0.0301,
      "step": 4087,
      "task_loss": 0.011213263496756554
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7863537541288032,
      "compression_loss": 0.0,
      "distillation_loss": 0.02990700677037239,
      "epoch": 3.88,
      "learning_rate": 2.0763768356744428e-05,
      "loss": 0.0282,
      "step": 4088,
      "task_loss": 0.012575274333357811
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7863631913662773,
      "compression_loss": 0.0,
      "distillation_loss": 0.042817167937755585,
      "epoch": 3.88,
      "learning_rate": 2.075326758253982e-05,
      "loss": 0.0542,
      "step": 4089,
      "task_loss": 0.15655487775802612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7863726242517859,
      "compression_loss": 0.0,
      "distillation_loss": 0.050644587725400925,
      "epoch": 3.88,
      "learning_rate": 2.0742767579774615e-05,
      "loss": 0.0495,
      "step": 4090,
      "task_loss": 0.03952382132411003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7863820527863326,
      "compression_loss": 0.0,
      "distillation_loss": 0.10132403671741486,
      "epoch": 3.89,
      "learning_rate": 2.073226835035618e-05,
      "loss": 0.105,
      "step": 4091,
      "task_loss": 0.13759130239486694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.786391476970921,
      "compression_loss": 0.0,
      "distillation_loss": 0.05861423537135124,
      "epoch": 3.89,
      "learning_rate": 2.0721769896191752e-05,
      "loss": 0.066,
      "step": 4092,
      "task_loss": 0.13297078013420105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7864008968065551,
      "compression_loss": 0.0,
      "distillation_loss": 0.02992338128387928,
      "epoch": 3.89,
      "learning_rate": 2.0711272219188423e-05,
      "loss": 0.0456,
      "step": 4093,
      "task_loss": 0.18631944060325623
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7864103122942382,
      "compression_loss": 0.0,
      "distillation_loss": 0.10860427469015121,
      "epoch": 3.89,
      "learning_rate": 2.0700775321253158e-05,
      "loss": 0.1143,
      "step": 4094,
      "task_loss": 0.16594167053699493
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7864197234349742,
      "compression_loss": 0.0,
      "distillation_loss": 0.08554205298423767,
      "epoch": 3.89,
      "learning_rate": 2.0690279204292753e-05,
      "loss": 0.1151,
      "step": 4095,
      "task_loss": 0.3806484341621399
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7864291302297668,
      "compression_loss": 0.0,
      "distillation_loss": 0.0441209115087986,
      "epoch": 3.89,
      "learning_rate": 2.0679783870213883e-05,
      "loss": 0.0408,
      "step": 4096,
      "task_loss": 0.011142965406179428
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7864385326796196,
      "compression_loss": 0.0,
      "distillation_loss": 0.05388110131025314,
      "epoch": 3.89,
      "learning_rate": 2.0669289320923086e-05,
      "loss": 0.049,
      "step": 4097,
      "task_loss": 0.004679244011640549
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7864479307855363,
      "compression_loss": 0.0,
      "distillation_loss": 0.033240221440792084,
      "epoch": 3.89,
      "learning_rate": 2.0658795558326743e-05,
      "loss": 0.031,
      "step": 4098,
      "task_loss": 0.011169865727424622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7864573245485206,
      "compression_loss": 0.0,
      "distillation_loss": 0.06061680614948273,
      "epoch": 3.89,
      "learning_rate": 2.0648302584331092e-05,
      "loss": 0.0627,
      "step": 4099,
      "task_loss": 0.08101128786802292
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7864667139695761,
      "compression_loss": 0.0,
      "distillation_loss": 0.12393200397491455,
      "epoch": 3.89,
      "learning_rate": 2.0637810400842233e-05,
      "loss": 0.1224,
      "step": 4100,
      "task_loss": 0.10840637236833572
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7864760990497066,
      "compression_loss": 0.0,
      "distillation_loss": 0.05604955554008484,
      "epoch": 3.89,
      "learning_rate": 2.0627319009766127e-05,
      "loss": 0.0707,
      "step": 4101,
      "task_loss": 0.20234975218772888
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7864854797899157,
      "compression_loss": 0.0,
      "distillation_loss": 0.03791045770049095,
      "epoch": 3.9,
      "learning_rate": 2.0616828413008578e-05,
      "loss": 0.0351,
      "step": 4102,
      "task_loss": 0.009459014981985092
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7864948561912071,
      "compression_loss": 0.0,
      "distillation_loss": 0.01791505143046379,
      "epoch": 3.9,
      "learning_rate": 2.0606338612475264e-05,
      "loss": 0.0255,
      "step": 4103,
      "task_loss": 0.09366492927074432
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7865042282545845,
      "compression_loss": 0.0,
      "distillation_loss": 0.14737428724765778,
      "epoch": 3.9,
      "learning_rate": 2.0595849610071697e-05,
      "loss": 0.1548,
      "step": 4104,
      "task_loss": 0.22147542238235474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7865135959810516,
      "compression_loss": 0.0,
      "distillation_loss": 0.04210761934518814,
      "epoch": 3.9,
      "learning_rate": 2.058536140770325e-05,
      "loss": 0.0384,
      "step": 4105,
      "task_loss": 0.004892520606517792
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7865229593716121,
      "compression_loss": 0.0,
      "distillation_loss": 0.03413204848766327,
      "epoch": 3.9,
      "learning_rate": 2.057487400727517e-05,
      "loss": 0.0316,
      "step": 4106,
      "task_loss": 0.008931750431656837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7865323184272696,
      "compression_loss": 0.0,
      "distillation_loss": 0.014508235268294811,
      "epoch": 3.9,
      "learning_rate": 2.0564387410692544e-05,
      "loss": 0.0136,
      "step": 4107,
      "task_loss": 0.00587865523993969
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7865416731490278,
      "compression_loss": 0.0,
      "distillation_loss": 0.1767938733100891,
      "epoch": 3.9,
      "learning_rate": 2.0553901619860306e-05,
      "loss": 0.1712,
      "step": 4108,
      "task_loss": 0.1211482584476471
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7865510235378904,
      "compression_loss": 0.0,
      "distillation_loss": 0.11559472978115082,
      "epoch": 3.9,
      "learning_rate": 2.0543416636683246e-05,
      "loss": 0.1049,
      "step": 4109,
      "task_loss": 0.008573394268751144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7865603695948611,
      "compression_loss": 0.0,
      "distillation_loss": 0.046687278896570206,
      "epoch": 3.9,
      "learning_rate": 2.0532932463066023e-05,
      "loss": 0.0461,
      "step": 4110,
      "task_loss": 0.04126298427581787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7865697113209436,
      "compression_loss": 0.0,
      "distillation_loss": 0.06493042409420013,
      "epoch": 3.9,
      "learning_rate": 2.052244910091313e-05,
      "loss": 0.0873,
      "step": 4111,
      "task_loss": 0.28819745779037476
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7865790487171415,
      "compression_loss": 0.0,
      "distillation_loss": 0.10494603216648102,
      "epoch": 3.91,
      "learning_rate": 2.051196655212892e-05,
      "loss": 0.0967,
      "step": 4112,
      "task_loss": 0.022337011992931366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7865883817844586,
      "compression_loss": 0.0,
      "distillation_loss": 0.12292291969060898,
      "epoch": 3.91,
      "learning_rate": 2.0501484818617594e-05,
      "loss": 0.1399,
      "step": 4113,
      "task_loss": 0.29296374320983887
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7865977105238985,
      "compression_loss": 0.0,
      "distillation_loss": 0.043324559926986694,
      "epoch": 3.91,
      "learning_rate": 2.0491003902283225e-05,
      "loss": 0.0394,
      "step": 4114,
      "task_loss": 0.004396416246891022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7866070349364648,
      "compression_loss": 0.0,
      "distillation_loss": 0.05673076957464218,
      "epoch": 3.91,
      "learning_rate": 2.048052380502971e-05,
      "loss": 0.0565,
      "step": 4115,
      "task_loss": 0.054648954421281815
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7866163550231614,
      "compression_loss": 0.0,
      "distillation_loss": 0.023540731519460678,
      "epoch": 3.91,
      "learning_rate": 2.047004452876081e-05,
      "loss": 0.0216,
      "step": 4116,
      "task_loss": 0.004315689206123352
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7866256707849918,
      "compression_loss": 0.0,
      "distillation_loss": 0.09178947657346725,
      "epoch": 3.91,
      "learning_rate": 2.045956607538015e-05,
      "loss": 0.0978,
      "step": 4117,
      "task_loss": 0.15143872797489166
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7866349822229597,
      "compression_loss": 0.0,
      "distillation_loss": 0.2266932725906372,
      "epoch": 3.91,
      "learning_rate": 2.0449088446791165e-05,
      "loss": 0.2216,
      "step": 4118,
      "task_loss": 0.17611849308013916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7866442893380688,
      "compression_loss": 0.0,
      "distillation_loss": 0.027620844542980194,
      "epoch": 3.91,
      "learning_rate": 2.043861164489719e-05,
      "loss": 0.0354,
      "step": 4119,
      "task_loss": 0.1055789515376091
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.786653592131323,
      "compression_loss": 0.0,
      "distillation_loss": 0.0539475753903389,
      "epoch": 3.91,
      "learning_rate": 2.0428135671601373e-05,
      "loss": 0.0601,
      "step": 4120,
      "task_loss": 0.11543746292591095
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7866628906037256,
      "compression_loss": 0.0,
      "distillation_loss": 0.047250404953956604,
      "epoch": 3.91,
      "learning_rate": 2.041766052880673e-05,
      "loss": 0.0593,
      "step": 4121,
      "task_loss": 0.167374387383461
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7866721847562806,
      "compression_loss": 0.0,
      "distillation_loss": 0.05512014403939247,
      "epoch": 3.91,
      "learning_rate": 2.0407186218416114e-05,
      "loss": 0.0648,
      "step": 4122,
      "task_loss": 0.15170332789421082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7866814745899915,
      "compression_loss": 0.0,
      "distillation_loss": 0.03062654659152031,
      "epoch": 3.92,
      "learning_rate": 2.039671274233225e-05,
      "loss": 0.0388,
      "step": 4123,
      "task_loss": 0.11201991140842438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.786690760105862,
      "compression_loss": 0.0,
      "distillation_loss": 0.01795804500579834,
      "epoch": 3.92,
      "learning_rate": 2.0386240102457682e-05,
      "loss": 0.0166,
      "step": 4124,
      "task_loss": 0.004647407680749893
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7867000413048959,
      "compression_loss": 0.0,
      "distillation_loss": 0.08352886140346527,
      "epoch": 3.92,
      "learning_rate": 2.0375768300694824e-05,
      "loss": 0.0783,
      "step": 4125,
      "task_loss": 0.030838267877697945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7867093181880968,
      "compression_loss": 0.0,
      "distillation_loss": 0.16173683106899261,
      "epoch": 3.92,
      "learning_rate": 2.0365297338945917e-05,
      "loss": 0.1537,
      "step": 4126,
      "task_loss": 0.08173392713069916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7867185907564682,
      "compression_loss": 0.0,
      "distillation_loss": 0.05559203773736954,
      "epoch": 3.92,
      "learning_rate": 2.035482721911308e-05,
      "loss": 0.0688,
      "step": 4127,
      "task_loss": 0.18805257976055145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7867278590110142,
      "compression_loss": 0.0,
      "distillation_loss": 0.03582768887281418,
      "epoch": 3.92,
      "learning_rate": 2.034435794309824e-05,
      "loss": 0.0388,
      "step": 4128,
      "task_loss": 0.06604026257991791
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7867371229527381,
      "compression_loss": 0.0,
      "distillation_loss": 0.042852409183979034,
      "epoch": 3.92,
      "learning_rate": 2.0333889512803204e-05,
      "loss": 0.0475,
      "step": 4129,
      "task_loss": 0.08887660503387451
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7867463825826438,
      "compression_loss": 0.0,
      "distillation_loss": 0.10272371768951416,
      "epoch": 3.92,
      "learning_rate": 2.0323421930129617e-05,
      "loss": 0.0932,
      "step": 4130,
      "task_loss": 0.007061410695314407
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7867556379017349,
      "compression_loss": 0.0,
      "distillation_loss": 0.15910032391548157,
      "epoch": 3.92,
      "learning_rate": 2.031295519697895e-05,
      "loss": 0.1638,
      "step": 4131,
      "task_loss": 0.2062944769859314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.786764888911015,
      "compression_loss": 0.0,
      "distillation_loss": 0.15020066499710083,
      "epoch": 3.92,
      "learning_rate": 2.0302489315252545e-05,
      "loss": 0.17,
      "step": 4132,
      "task_loss": 0.3483584523200989
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.786774135611488,
      "compression_loss": 0.0,
      "distillation_loss": 0.07658859342336655,
      "epoch": 3.92,
      "learning_rate": 2.0292024286851584e-05,
      "loss": 0.0723,
      "step": 4133,
      "task_loss": 0.03387470170855522
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7867833780041574,
      "compression_loss": 0.0,
      "distillation_loss": 0.06731578707695007,
      "epoch": 3.93,
      "learning_rate": 2.0281560113677086e-05,
      "loss": 0.0683,
      "step": 4134,
      "task_loss": 0.07722228765487671
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7867926160900269,
      "compression_loss": 0.0,
      "distillation_loss": 0.03083537518978119,
      "epoch": 3.93,
      "learning_rate": 2.0271096797629915e-05,
      "loss": 0.0292,
      "step": 4135,
      "task_loss": 0.014749417081475258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7868018498701004,
      "compression_loss": 0.0,
      "distillation_loss": 0.055131688714027405,
      "epoch": 3.93,
      "learning_rate": 2.0260634340610786e-05,
      "loss": 0.0589,
      "step": 4136,
      "task_loss": 0.09324628859758377
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7868110793453812,
      "compression_loss": 0.0,
      "distillation_loss": 0.06215892732143402,
      "epoch": 3.93,
      "learning_rate": 2.0250172744520258e-05,
      "loss": 0.0583,
      "step": 4137,
      "task_loss": 0.023460306227207184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7868203045168732,
      "compression_loss": 0.0,
      "distillation_loss": 0.11791975051164627,
      "epoch": 3.93,
      "learning_rate": 2.023971201125872e-05,
      "loss": 0.1233,
      "step": 4138,
      "task_loss": 0.17152175307273865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7868295253855802,
      "compression_loss": 0.0,
      "distillation_loss": 0.03688070923089981,
      "epoch": 3.93,
      "learning_rate": 2.0229252142726415e-05,
      "loss": 0.0461,
      "step": 4139,
      "task_loss": 0.12953397631645203
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7868387419525057,
      "compression_loss": 0.0,
      "distillation_loss": 0.059014152735471725,
      "epoch": 3.93,
      "learning_rate": 2.021879314082344e-05,
      "loss": 0.0663,
      "step": 4140,
      "task_loss": 0.13181355595588684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7868479542186534,
      "compression_loss": 0.0,
      "distillation_loss": 0.31541213393211365,
      "epoch": 3.93,
      "learning_rate": 2.0208335007449726e-05,
      "loss": 0.3036,
      "step": 4141,
      "task_loss": 0.19743433594703674
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.786857162185027,
      "compression_loss": 0.0,
      "distillation_loss": 0.029475336894392967,
      "epoch": 3.93,
      "learning_rate": 2.019787774450503e-05,
      "loss": 0.0269,
      "step": 4142,
      "task_loss": 0.004132760688662529
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7868663658526301,
      "compression_loss": 0.0,
      "distillation_loss": 0.13314858078956604,
      "epoch": 3.93,
      "learning_rate": 2.0187421353888966e-05,
      "loss": 0.1294,
      "step": 4143,
      "task_loss": 0.09587833285331726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7868755652224667,
      "compression_loss": 0.0,
      "distillation_loss": 0.05543944984674454,
      "epoch": 3.94,
      "learning_rate": 2.0176965837500995e-05,
      "loss": 0.0528,
      "step": 4144,
      "task_loss": 0.029170924797654152
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7868847602955401,
      "compression_loss": 0.0,
      "distillation_loss": 0.031388137489557266,
      "epoch": 3.94,
      "learning_rate": 2.0166511197240405e-05,
      "loss": 0.0293,
      "step": 4145,
      "task_loss": 0.01011231541633606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7868939510728542,
      "compression_loss": 0.0,
      "distillation_loss": 0.1392134577035904,
      "epoch": 3.94,
      "learning_rate": 2.015605743500634e-05,
      "loss": 0.1419,
      "step": 4146,
      "task_loss": 0.16592922806739807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7869031375554126,
      "compression_loss": 0.0,
      "distillation_loss": 0.04586111381649971,
      "epoch": 3.94,
      "learning_rate": 2.0145604552697763e-05,
      "loss": 0.0487,
      "step": 4147,
      "task_loss": 0.07397586107254028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.786912319744219,
      "compression_loss": 0.0,
      "distillation_loss": 0.030558858066797256,
      "epoch": 3.94,
      "learning_rate": 2.0135152552213493e-05,
      "loss": 0.0542,
      "step": 4148,
      "task_loss": 0.2674226462841034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7869214976402772,
      "compression_loss": 0.0,
      "distillation_loss": 0.1152239590883255,
      "epoch": 3.94,
      "learning_rate": 2.0124701435452198e-05,
      "loss": 0.1043,
      "step": 4149,
      "task_loss": 0.005567222833633423
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7869306712445906,
      "compression_loss": 0.0,
      "distillation_loss": 0.15521375834941864,
      "epoch": 3.94,
      "learning_rate": 2.0114251204312367e-05,
      "loss": 0.1738,
      "step": 4150,
      "task_loss": 0.3411465287208557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7869398405581631,
      "compression_loss": 0.0,
      "distillation_loss": 0.07683128863573074,
      "epoch": 3.94,
      "learning_rate": 2.010380186069234e-05,
      "loss": 0.0698,
      "step": 4151,
      "task_loss": 0.0065435003489255905
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7869490055819984,
      "compression_loss": 0.0,
      "distillation_loss": 0.07263028621673584,
      "epoch": 3.94,
      "learning_rate": 2.009335340649028e-05,
      "loss": 0.0779,
      "step": 4152,
      "task_loss": 0.12566472589969635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7869581663171001,
      "compression_loss": 0.0,
      "distillation_loss": 0.10479018092155457,
      "epoch": 3.94,
      "learning_rate": 2.0082905843604206e-05,
      "loss": 0.1073,
      "step": 4153,
      "task_loss": 0.13012909889221191
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7869673227644719,
      "compression_loss": 0.0,
      "distillation_loss": 0.0787314772605896,
      "epoch": 3.94,
      "learning_rate": 2.0072459173931964e-05,
      "loss": 0.0745,
      "step": 4154,
      "task_loss": 0.03596143424510956
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7869764749251175,
      "compression_loss": 0.0,
      "distillation_loss": 0.02369462139904499,
      "epoch": 3.95,
      "learning_rate": 2.0062013399371245e-05,
      "loss": 0.022,
      "step": 4155,
      "task_loss": 0.007038429379463196
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7869856228000405,
      "compression_loss": 0.0,
      "distillation_loss": 0.018353287130594254,
      "epoch": 3.95,
      "learning_rate": 2.005156852181958e-05,
      "loss": 0.0253,
      "step": 4156,
      "task_loss": 0.08732615411281586
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7869947663902447,
      "compression_loss": 0.0,
      "distillation_loss": 0.026407258585095406,
      "epoch": 3.95,
      "learning_rate": 2.004112454317431e-05,
      "loss": 0.0248,
      "step": 4157,
      "task_loss": 0.010559692978858948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7870039056967336,
      "compression_loss": 0.0,
      "distillation_loss": 0.018740851432085037,
      "epoch": 3.95,
      "learning_rate": 2.003068146533266e-05,
      "loss": 0.0175,
      "step": 4158,
      "task_loss": 0.00584782287478447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7870130407205111,
      "compression_loss": 0.0,
      "distillation_loss": 0.1747373789548874,
      "epoch": 3.95,
      "learning_rate": 2.002023929019165e-05,
      "loss": 0.1671,
      "step": 4159,
      "task_loss": 0.09856338798999786
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7870221714625809,
      "compression_loss": 0.0,
      "distillation_loss": 0.05840107798576355,
      "epoch": 3.95,
      "learning_rate": 2.0009798019648163e-05,
      "loss": 0.0601,
      "step": 4160,
      "task_loss": 0.07528725266456604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7870312979239464,
      "compression_loss": 0.0,
      "distillation_loss": 0.052719514816999435,
      "epoch": 3.95,
      "learning_rate": 1.9999357655598893e-05,
      "loss": 0.0555,
      "step": 4161,
      "task_loss": 0.08041465282440186
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7870404201056116,
      "compression_loss": 0.0,
      "distillation_loss": 0.11436031758785248,
      "epoch": 3.95,
      "learning_rate": 1.9988918199940386e-05,
      "loss": 0.1258,
      "step": 4162,
      "task_loss": 0.22912541031837463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7870495380085799,
      "compression_loss": 0.0,
      "distillation_loss": 0.027497505769133568,
      "epoch": 3.95,
      "learning_rate": 1.997847965456903e-05,
      "loss": 0.0335,
      "step": 4163,
      "task_loss": 0.08756634593009949
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7870586516338552,
      "compression_loss": 0.0,
      "distillation_loss": 0.12368052452802658,
      "epoch": 3.95,
      "learning_rate": 1.9968042021381023e-05,
      "loss": 0.1237,
      "step": 4164,
      "task_loss": 0.12429526448249817
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7870677609824411,
      "compression_loss": 0.0,
      "distillation_loss": 0.03644444793462753,
      "epoch": 3.96,
      "learning_rate": 1.9957605302272412e-05,
      "loss": 0.0463,
      "step": 4165,
      "task_loss": 0.13482612371444702
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7870768660553412,
      "compression_loss": 0.0,
      "distillation_loss": 0.1511649489402771,
      "epoch": 3.96,
      "learning_rate": 1.9947169499139083e-05,
      "loss": 0.1411,
      "step": 4166,
      "task_loss": 0.05041804164648056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7870859668535594,
      "compression_loss": 0.0,
      "distillation_loss": 0.16785401105880737,
      "epoch": 3.96,
      "learning_rate": 1.9936734613876762e-05,
      "loss": 0.1752,
      "step": 4167,
      "task_loss": 0.24086761474609375
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7870950633780992,
      "compression_loss": 0.0,
      "distillation_loss": 0.06589843332767487,
      "epoch": 3.96,
      "learning_rate": 1.9926300648380976e-05,
      "loss": 0.0679,
      "step": 4168,
      "task_loss": 0.08612485975027084
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7871041556299643,
      "compression_loss": 0.0,
      "distillation_loss": 0.1752784252166748,
      "epoch": 3.96,
      "learning_rate": 1.9915867604547106e-05,
      "loss": 0.168,
      "step": 4169,
      "task_loss": 0.10218894481658936
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7871132436101584,
      "compression_loss": 0.0,
      "distillation_loss": 0.06247349828481674,
      "epoch": 3.96,
      "learning_rate": 1.9905435484270383e-05,
      "loss": 0.0576,
      "step": 4170,
      "task_loss": 0.014096638187766075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7871223273196852,
      "compression_loss": 0.0,
      "distillation_loss": 0.032126858830451965,
      "epoch": 3.96,
      "learning_rate": 1.989500428944583e-05,
      "loss": 0.0376,
      "step": 4171,
      "task_loss": 0.08727513253688812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7871314067595484,
      "compression_loss": 0.0,
      "distillation_loss": 0.032654955983161926,
      "epoch": 3.96,
      "learning_rate": 1.988457402196834e-05,
      "loss": 0.0306,
      "step": 4172,
      "task_loss": 0.01211773045361042
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7871404819307517,
      "compression_loss": 0.0,
      "distillation_loss": 0.018787803128361702,
      "epoch": 3.96,
      "learning_rate": 1.9874144683732615e-05,
      "loss": 0.0173,
      "step": 4173,
      "task_loss": 0.0037378836423158646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7871495528342987,
      "compression_loss": 0.0,
      "distillation_loss": 0.04686371237039566,
      "epoch": 3.96,
      "learning_rate": 1.9863716276633186e-05,
      "loss": 0.0444,
      "step": 4174,
      "task_loss": 0.021781545132398605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.787158619471193,
      "compression_loss": 0.0,
      "distillation_loss": 0.089274100959301,
      "epoch": 3.96,
      "learning_rate": 1.9853288802564438e-05,
      "loss": 0.0895,
      "step": 4175,
      "task_loss": 0.09111557900905609
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7871676818424386,
      "compression_loss": 0.0,
      "distillation_loss": 0.0693182647228241,
      "epoch": 3.97,
      "learning_rate": 1.9842862263420564e-05,
      "loss": 0.0683,
      "step": 4176,
      "task_loss": 0.05941478908061981
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7871767399490389,
      "compression_loss": 0.0,
      "distillation_loss": 0.02392822504043579,
      "epoch": 3.97,
      "learning_rate": 1.9832436661095604e-05,
      "loss": 0.0366,
      "step": 4177,
      "task_loss": 0.15036530792713165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7871857937919977,
      "compression_loss": 0.0,
      "distillation_loss": 0.07844184339046478,
      "epoch": 3.97,
      "learning_rate": 1.98220119974834e-05,
      "loss": 0.0762,
      "step": 4178,
      "task_loss": 0.0561903640627861
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7871948433723186,
      "compression_loss": 0.0,
      "distillation_loss": 0.08809831738471985,
      "epoch": 3.97,
      "learning_rate": 1.9811588274477665e-05,
      "loss": 0.0907,
      "step": 4179,
      "task_loss": 0.11436465382575989
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7872038886910053,
      "compression_loss": 0.0,
      "distillation_loss": 0.02092152088880539,
      "epoch": 3.97,
      "learning_rate": 1.980116549397191e-05,
      "loss": 0.0242,
      "step": 4180,
      "task_loss": 0.05349248647689819
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7872129297490617,
      "compression_loss": 0.0,
      "distillation_loss": 0.10855552554130554,
      "epoch": 3.97,
      "learning_rate": 1.979074365785947e-05,
      "loss": 0.104,
      "step": 4181,
      "task_loss": 0.06298065930604935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7872219665474911,
      "compression_loss": 0.0,
      "distillation_loss": 0.07886943221092224,
      "epoch": 3.97,
      "learning_rate": 1.978032276803354e-05,
      "loss": 0.0749,
      "step": 4182,
      "task_loss": 0.03924179822206497
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7872309990872974,
      "compression_loss": 0.0,
      "distillation_loss": 0.024403858929872513,
      "epoch": 3.97,
      "learning_rate": 1.9769902826387106e-05,
      "loss": 0.0229,
      "step": 4183,
      "task_loss": 0.008915219455957413
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7872400273694844,
      "compression_loss": 0.0,
      "distillation_loss": 0.08649304509162903,
      "epoch": 3.97,
      "learning_rate": 1.9759483834813023e-05,
      "loss": 0.0885,
      "step": 4184,
      "task_loss": 0.10659407079219818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7872490513950555,
      "compression_loss": 0.0,
      "distillation_loss": 0.030071692541241646,
      "epoch": 3.97,
      "learning_rate": 1.9749065795203938e-05,
      "loss": 0.0468,
      "step": 4185,
      "task_loss": 0.19697120785713196
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7872580711650146,
      "compression_loss": 0.0,
      "distillation_loss": 0.05038078874349594,
      "epoch": 3.98,
      "learning_rate": 1.9738648709452336e-05,
      "loss": 0.0527,
      "step": 4186,
      "task_loss": 0.07359597086906433
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7872670866803653,
      "compression_loss": 0.0,
      "distillation_loss": 0.09490829706192017,
      "epoch": 3.98,
      "learning_rate": 1.9728232579450543e-05,
      "loss": 0.1038,
      "step": 4187,
      "task_loss": 0.18415942788124084
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7872760979421113,
      "compression_loss": 0.0,
      "distillation_loss": 0.1876422017812729,
      "epoch": 3.98,
      "learning_rate": 1.971781740709068e-05,
      "loss": 0.1938,
      "step": 4188,
      "task_loss": 0.24940979480743408
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7872851049512563,
      "compression_loss": 0.0,
      "distillation_loss": 0.021935122087597847,
      "epoch": 3.98,
      "learning_rate": 1.970740319426474e-05,
      "loss": 0.0206,
      "step": 4189,
      "task_loss": 0.008345887064933777
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7872941077088038,
      "compression_loss": 0.0,
      "distillation_loss": 0.07562540471553802,
      "epoch": 3.98,
      "learning_rate": 1.9696989942864488e-05,
      "loss": 0.0898,
      "step": 4190,
      "task_loss": 0.21687836945056915
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7873031062157578,
      "compression_loss": 0.0,
      "distillation_loss": 0.06813269853591919,
      "epoch": 3.98,
      "learning_rate": 1.9686577654781546e-05,
      "loss": 0.0756,
      "step": 4191,
      "task_loss": 0.14326757192611694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7873121004731217,
      "compression_loss": 0.0,
      "distillation_loss": 0.034589093178510666,
      "epoch": 3.98,
      "learning_rate": 1.967616633190737e-05,
      "loss": 0.0422,
      "step": 4192,
      "task_loss": 0.11064015328884125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7873210904818994,
      "compression_loss": 0.0,
      "distillation_loss": 0.020227717235684395,
      "epoch": 3.98,
      "learning_rate": 1.966575597613322e-05,
      "loss": 0.0186,
      "step": 4193,
      "task_loss": 0.004256442189216614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7873300762430945,
      "compression_loss": 0.0,
      "distillation_loss": 0.10587961226701736,
      "epoch": 3.98,
      "learning_rate": 1.9655346589350194e-05,
      "loss": 0.1102,
      "step": 4194,
      "task_loss": 0.14888451993465424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7873390577577105,
      "compression_loss": 0.0,
      "distillation_loss": 0.08755764365196228,
      "epoch": 3.98,
      "learning_rate": 1.964493817344919e-05,
      "loss": 0.1018,
      "step": 4195,
      "task_loss": 0.22955255210399628
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7873480350267514,
      "compression_loss": 0.0,
      "distillation_loss": 0.16795973479747772,
      "epoch": 3.98,
      "learning_rate": 1.9634530730320967e-05,
      "loss": 0.1533,
      "step": 4196,
      "task_loss": 0.02090776339173317
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7873570080512207,
      "compression_loss": 0.0,
      "distillation_loss": 0.1484326869249344,
      "epoch": 3.99,
      "learning_rate": 1.9624124261856068e-05,
      "loss": 0.1569,
      "step": 4197,
      "task_loss": 0.23264384269714355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7873659768321221,
      "compression_loss": 0.0,
      "distillation_loss": 0.07327691465616226,
      "epoch": 3.99,
      "learning_rate": 1.961371876994489e-05,
      "loss": 0.0931,
      "step": 4198,
      "task_loss": 0.2711363434791565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7873749413704593,
      "compression_loss": 0.0,
      "distillation_loss": 0.04226084426045418,
      "epoch": 3.99,
      "learning_rate": 1.9603314256477644e-05,
      "loss": 0.0453,
      "step": 4199,
      "task_loss": 0.0727810189127922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7873839016672359,
      "compression_loss": 0.0,
      "distillation_loss": 0.1644660234451294,
      "epoch": 3.99,
      "learning_rate": 1.9592910723344335e-05,
      "loss": 0.1607,
      "step": 4200,
      "task_loss": 0.12694597244262695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7873928577234557,
      "compression_loss": 0.0,
      "distillation_loss": 0.145688995718956,
      "epoch": 3.99,
      "learning_rate": 1.958250817243485e-05,
      "loss": 0.1528,
      "step": 4201,
      "task_loss": 0.2163536101579666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7874018095401224,
      "compression_loss": 0.0,
      "distillation_loss": 0.04844709113240242,
      "epoch": 3.99,
      "learning_rate": 1.9572106605638842e-05,
      "loss": 0.0536,
      "step": 4202,
      "task_loss": 0.09967800229787827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7874107571182396,
      "compression_loss": 0.0,
      "distillation_loss": 0.019720932468771935,
      "epoch": 3.99,
      "learning_rate": 1.956170602484582e-05,
      "loss": 0.0182,
      "step": 4203,
      "task_loss": 0.004351753741502762
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7874197004588109,
      "compression_loss": 0.0,
      "distillation_loss": 0.10592324286699295,
      "epoch": 3.99,
      "learning_rate": 1.955130643194508e-05,
      "loss": 0.1063,
      "step": 4204,
      "task_loss": 0.10958631336688995
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7874286395628401,
      "compression_loss": 0.0,
      "distillation_loss": 0.05713961645960808,
      "epoch": 3.99,
      "learning_rate": 1.9540907828825768e-05,
      "loss": 0.0603,
      "step": 4205,
      "task_loss": 0.08864769339561462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.787437574431331,
      "compression_loss": 0.0,
      "distillation_loss": 0.0976656973361969,
      "epoch": 3.99,
      "learning_rate": 1.9530510217376843e-05,
      "loss": 0.1037,
      "step": 4206,
      "task_loss": 0.15841828286647797
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7874465050652871,
      "compression_loss": 0.0,
      "distillation_loss": 0.13762739300727844,
      "epoch": 4.0,
      "learning_rate": 1.952011359948708e-05,
      "loss": 0.1429,
      "step": 4207,
      "task_loss": 0.1902618259191513
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7874554314657122,
      "compression_loss": 0.0,
      "distillation_loss": 0.047601085156202316,
      "epoch": 4.0,
      "learning_rate": 1.9509717977045068e-05,
      "loss": 0.0439,
      "step": 4208,
      "task_loss": 0.01073162630200386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7874643536336098,
      "compression_loss": 0.0,
      "distillation_loss": 0.056678276509046555,
      "epoch": 4.0,
      "learning_rate": 1.949932335193922e-05,
      "loss": 0.0562,
      "step": 4209,
      "task_loss": 0.052305128425359726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7874732715699837,
      "compression_loss": 0.0,
      "distillation_loss": 0.18532370030879974,
      "epoch": 4.0,
      "learning_rate": 1.948892972605779e-05,
      "loss": 0.1904,
      "step": 4210,
      "task_loss": 0.2363312840461731
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7874821852758376,
      "compression_loss": 0.0,
      "distillation_loss": 0.1466386467218399,
      "epoch": 4.0,
      "learning_rate": 1.9478537101288814e-05,
      "loss": 0.1392,
      "step": 4211,
      "task_loss": 0.07213930040597916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6058685241936025,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7755859339922954,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7874910947521752,
      "compression_loss": 0.0,
      "distillation_loss": 0.078835628926754,
      "epoch": 4.0,
      "learning_rate": 1.946814547952016e-05,
      "loss": 0.0736,
      "step": 4212,
      "task_loss": 0.02693953923881054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7875000000000001,
      "compression_loss": 0.0,
      "distillation_loss": 0.4135439693927765,
      "epoch": 4.0,
      "learning_rate": 1.945775486263953e-05,
      "loss": 0.3929,
      "step": 4213,
      "task_loss": 0.20665842294692993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.787508901020316,
      "compression_loss": 0.0,
      "distillation_loss": 0.32134756445884705,
      "epoch": 4.0,
      "learning_rate": 1.9447365252534414e-05,
      "loss": 0.2996,
      "step": 4214,
      "task_loss": 0.10436099767684937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7875177978141266,
      "compression_loss": 0.0,
      "distillation_loss": 0.3447112441062927,
      "epoch": 4.0,
      "learning_rate": 1.9436976651092144e-05,
      "loss": 0.3183,
      "step": 4215,
      "task_loss": 0.08028891682624817
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7875266903824356,
      "compression_loss": 0.0,
      "distillation_loss": 0.2967658042907715,
      "epoch": 4.0,
      "learning_rate": 1.942658906019986e-05,
      "loss": 0.2787,
      "step": 4216,
      "task_loss": 0.11631828546524048
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7875355787262467,
      "compression_loss": 0.0,
      "distillation_loss": 0.3276255130767822,
      "epoch": 4.0,
      "learning_rate": 1.9416202481744504e-05,
      "loss": 0.3027,
      "step": 4217,
      "task_loss": 0.07845309376716614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7875444628465635,
      "compression_loss": 0.0,
      "distillation_loss": 0.3934395909309387,
      "epoch": 4.01,
      "learning_rate": 1.940581691761287e-05,
      "loss": 0.3804,
      "step": 4218,
      "task_loss": 0.2632703185081482
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7875533427443898,
      "compression_loss": 0.0,
      "distillation_loss": 0.3027340769767761,
      "epoch": 4.01,
      "learning_rate": 1.9395432369691526e-05,
      "loss": 0.2816,
      "step": 4219,
      "task_loss": 0.09113991260528564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7875622184207293,
      "compression_loss": 0.0,
      "distillation_loss": 0.30732953548431396,
      "epoch": 4.01,
      "learning_rate": 1.9385048839866896e-05,
      "loss": 0.2857,
      "step": 4220,
      "task_loss": 0.09076324105262756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7875710898765854,
      "compression_loss": 0.0,
      "distillation_loss": 0.27244919538497925,
      "epoch": 4.01,
      "learning_rate": 1.9374666330025178e-05,
      "loss": 0.26,
      "step": 4221,
      "task_loss": 0.14839491248130798
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7875799571129621,
      "compression_loss": 0.0,
      "distillation_loss": 0.13425913453102112,
      "epoch": 4.01,
      "learning_rate": 1.9364284842052414e-05,
      "loss": 0.1228,
      "step": 4222,
      "task_loss": 0.019507169723510742
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.787588820130863,
      "compression_loss": 0.0,
      "distillation_loss": 0.17389854788780212,
      "epoch": 4.01,
      "learning_rate": 1.9353904377834454e-05,
      "loss": 0.1588,
      "step": 4223,
      "task_loss": 0.0224696546792984
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7875976789312916,
      "compression_loss": 0.0,
      "distillation_loss": 0.16582781076431274,
      "epoch": 4.01,
      "learning_rate": 1.934352493925695e-05,
      "loss": 0.1547,
      "step": 4224,
      "task_loss": 0.05480484664440155
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7876065335152518,
      "compression_loss": 0.0,
      "distillation_loss": 0.09588154405355453,
      "epoch": 4.01,
      "learning_rate": 1.933314652820539e-05,
      "loss": 0.0882,
      "step": 4225,
      "task_loss": 0.019207999110221863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7876153838837472,
      "compression_loss": 0.0,
      "distillation_loss": 0.11465960741043091,
      "epoch": 4.01,
      "learning_rate": 1.932276914656504e-05,
      "loss": 0.106,
      "step": 4226,
      "task_loss": 0.028103739023208618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7876242300377816,
      "compression_loss": 0.0,
      "distillation_loss": 0.22152969241142273,
      "epoch": 4.01,
      "learning_rate": 1.9312392796221033e-05,
      "loss": 0.217,
      "step": 4227,
      "task_loss": 0.1767098605632782
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7876330719783584,
      "compression_loss": 0.0,
      "distillation_loss": 0.17856980860233307,
      "epoch": 4.02,
      "learning_rate": 1.9302017479058256e-05,
      "loss": 0.1757,
      "step": 4228,
      "task_loss": 0.149837926030159
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7876419097064816,
      "compression_loss": 0.0,
      "distillation_loss": 0.08909793198108673,
      "epoch": 4.02,
      "learning_rate": 1.929164319696145e-05,
      "loss": 0.09,
      "step": 4229,
      "task_loss": 0.09795405715703964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7876507432231546,
      "compression_loss": 0.0,
      "distillation_loss": 0.11065828800201416,
      "epoch": 4.02,
      "learning_rate": 1.9281269951815154e-05,
      "loss": 0.1068,
      "step": 4230,
      "task_loss": 0.07157891988754272
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7876595725293813,
      "compression_loss": 0.0,
      "distillation_loss": 0.1124470978975296,
      "epoch": 4.02,
      "learning_rate": 1.9270897745503706e-05,
      "loss": 0.107,
      "step": 4231,
      "task_loss": 0.05780310928821564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7876683976261654,
      "compression_loss": 0.0,
      "distillation_loss": 0.09072036296129227,
      "epoch": 4.02,
      "learning_rate": 1.9260526579911283e-05,
      "loss": 0.0887,
      "step": 4232,
      "task_loss": 0.07040438801050186
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7876772185145103,
      "compression_loss": 0.0,
      "distillation_loss": 0.1262628436088562,
      "epoch": 4.02,
      "learning_rate": 1.9250156456921837e-05,
      "loss": 0.1201,
      "step": 4233,
      "task_loss": 0.06450807303190231
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.78768603519542,
      "compression_loss": 0.0,
      "distillation_loss": 0.062325261533260345,
      "epoch": 4.02,
      "learning_rate": 1.9239787378419165e-05,
      "loss": 0.0646,
      "step": 4234,
      "task_loss": 0.08547300100326538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.787694847669898,
      "compression_loss": 0.0,
      "distillation_loss": 0.054038092494010925,
      "epoch": 4.02,
      "learning_rate": 1.9229419346286853e-05,
      "loss": 0.0545,
      "step": 4235,
      "task_loss": 0.05840130150318146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.787703655938948,
      "compression_loss": 0.0,
      "distillation_loss": 0.11831286549568176,
      "epoch": 4.02,
      "learning_rate": 1.9219052362408314e-05,
      "loss": 0.1204,
      "step": 4236,
      "task_loss": 0.13891497254371643
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7877124600035738,
      "compression_loss": 0.0,
      "distillation_loss": 0.13940417766571045,
      "epoch": 4.02,
      "learning_rate": 1.920868642866676e-05,
      "loss": 0.1331,
      "step": 4237,
      "task_loss": 0.07673609256744385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7877212598647789,
      "compression_loss": 0.0,
      "distillation_loss": 0.11010223627090454,
      "epoch": 4.02,
      "learning_rate": 1.91983215469452e-05,
      "loss": 0.1152,
      "step": 4238,
      "task_loss": 0.16106702387332916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7877300555235671,
      "compression_loss": 0.0,
      "distillation_loss": 0.18333761394023895,
      "epoch": 4.03,
      "learning_rate": 1.918795771912648e-05,
      "loss": 0.1795,
      "step": 4239,
      "task_loss": 0.14447200298309326
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7877388469809421,
      "compression_loss": 0.0,
      "distillation_loss": 0.08234288543462753,
      "epoch": 4.03,
      "learning_rate": 1.917759494709322e-05,
      "loss": 0.0861,
      "step": 4240,
      "task_loss": 0.12013165652751923
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7877476342379075,
      "compression_loss": 0.0,
      "distillation_loss": 0.134935200214386,
      "epoch": 4.03,
      "learning_rate": 1.9167233232727885e-05,
      "loss": 0.1334,
      "step": 4241,
      "task_loss": 0.11945871263742447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.787756417295467,
      "compression_loss": 0.0,
      "distillation_loss": 0.053517088294029236,
      "epoch": 4.03,
      "learning_rate": 1.915687257791273e-05,
      "loss": 0.0582,
      "step": 4242,
      "task_loss": 0.10031777620315552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7877651961546244,
      "compression_loss": 0.0,
      "distillation_loss": 0.1066487580537796,
      "epoch": 4.03,
      "learning_rate": 1.9146512984529793e-05,
      "loss": 0.1059,
      "step": 4243,
      "task_loss": 0.09886283427476883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7877739708163831,
      "compression_loss": 0.0,
      "distillation_loss": 0.23031002283096313,
      "epoch": 4.03,
      "learning_rate": 1.913615445446098e-05,
      "loss": 0.2218,
      "step": 4244,
      "task_loss": 0.14517052471637726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7877827412817472,
      "compression_loss": 0.0,
      "distillation_loss": 0.2954586148262024,
      "epoch": 4.03,
      "learning_rate": 1.9125796989587947e-05,
      "loss": 0.3012,
      "step": 4245,
      "task_loss": 0.35336050391197205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.78779150755172,
      "compression_loss": 0.0,
      "distillation_loss": 0.0650944635272026,
      "epoch": 4.03,
      "learning_rate": 1.9115440591792182e-05,
      "loss": 0.063,
      "step": 4246,
      "task_loss": 0.044294241815805435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7878002696273054,
      "compression_loss": 0.0,
      "distillation_loss": 0.070266954600811,
      "epoch": 4.03,
      "learning_rate": 1.9105085262954975e-05,
      "loss": 0.0739,
      "step": 4247,
      "task_loss": 0.10610578209161758
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.787809027509507,
      "compression_loss": 0.0,
      "distillation_loss": 0.09126845002174377,
      "epoch": 4.03,
      "learning_rate": 1.9094731004957416e-05,
      "loss": 0.0977,
      "step": 4248,
      "task_loss": 0.15584436058998108
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7878177811993284,
      "compression_loss": 0.0,
      "distillation_loss": 0.1306382119655609,
      "epoch": 4.04,
      "learning_rate": 1.9084377819680417e-05,
      "loss": 0.1219,
      "step": 4249,
      "task_loss": 0.043007947504520416
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7878265306977735,
      "compression_loss": 0.0,
      "distillation_loss": 0.06510676443576813,
      "epoch": 4.04,
      "learning_rate": 1.9074025709004672e-05,
      "loss": 0.0593,
      "step": 4250,
      "task_loss": 0.007219014689326286
    },
    {
      "epoch": 4.04,
      "eval_accuracy": 0.8910550458715596,
      "eval_loss": 0.53865647315979,
      "eval_runtime": 17.9464,
      "eval_samples_per_second": 48.589,
      "eval_steps_per_second": 6.074,
      "step": 4250
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7878352760058459,
      "compression_loss": 0.0,
      "distillation_loss": 0.07825392484664917,
      "epoch": 4.04,
      "learning_rate": 1.9063674674810696e-05,
      "loss": 0.0739,
      "step": 4251,
      "task_loss": 0.035188619047403336
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7878440171245491,
      "compression_loss": 0.0,
      "distillation_loss": 0.1569930762052536,
      "epoch": 4.04,
      "learning_rate": 1.90533247189788e-05,
      "loss": 0.1545,
      "step": 4252,
      "task_loss": 0.1325351744890213
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7878527540548871,
      "compression_loss": 0.0,
      "distillation_loss": 0.0776907354593277,
      "epoch": 4.04,
      "learning_rate": 1.9042975843389115e-05,
      "loss": 0.0752,
      "step": 4253,
      "task_loss": 0.05311368405818939
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7878614867978633,
      "compression_loss": 0.0,
      "distillation_loss": 0.02283748984336853,
      "epoch": 4.04,
      "learning_rate": 1.903262804992156e-05,
      "loss": 0.0229,
      "step": 4254,
      "task_loss": 0.023530958220362663
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7878702153544814,
      "compression_loss": 0.0,
      "distillation_loss": 0.35799717903137207,
      "epoch": 4.04,
      "learning_rate": 1.9022281340455854e-05,
      "loss": 0.3416,
      "step": 4255,
      "task_loss": 0.1938759684562683
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7878789397257453,
      "compression_loss": 0.0,
      "distillation_loss": 0.08360302448272705,
      "epoch": 4.04,
      "learning_rate": 1.9011935716871535e-05,
      "loss": 0.0995,
      "step": 4256,
      "task_loss": 0.24218979477882385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7878876599126585,
      "compression_loss": 0.0,
      "distillation_loss": 0.060659270733594894,
      "epoch": 4.04,
      "learning_rate": 1.900159118104793e-05,
      "loss": 0.0644,
      "step": 4257,
      "task_loss": 0.09822718799114227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7878963759162247,
      "compression_loss": 0.0,
      "distillation_loss": 0.10400435328483582,
      "epoch": 4.04,
      "learning_rate": 1.8991247734864173e-05,
      "loss": 0.1054,
      "step": 4258,
      "task_loss": 0.11802786588668823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7879050877374478,
      "compression_loss": 0.0,
      "distillation_loss": 0.12243454158306122,
      "epoch": 4.04,
      "learning_rate": 1.898090538019921e-05,
      "loss": 0.1253,
      "step": 4259,
      "task_loss": 0.15127399563789368
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7879137953773312,
      "compression_loss": 0.0,
      "distillation_loss": 0.04609953239560127,
      "epoch": 4.05,
      "learning_rate": 1.897056411893177e-05,
      "loss": 0.0505,
      "step": 4260,
      "task_loss": 0.08963650465011597
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7879224988368786,
      "compression_loss": 0.0,
      "distillation_loss": 0.02292802557349205,
      "epoch": 4.05,
      "learning_rate": 1.896022395294039e-05,
      "loss": 0.0211,
      "step": 4261,
      "task_loss": 0.004292329773306847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7879311981170939,
      "compression_loss": 0.0,
      "distillation_loss": 0.037770919501781464,
      "epoch": 4.05,
      "learning_rate": 1.8949884884103418e-05,
      "loss": 0.0359,
      "step": 4262,
      "task_loss": 0.019271956756711006
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7879398932189805,
      "compression_loss": 0.0,
      "distillation_loss": 0.16756854951381683,
      "epoch": 4.05,
      "learning_rate": 1.8939546914299e-05,
      "loss": 0.171,
      "step": 4263,
      "task_loss": 0.20192977786064148
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7879485841435423,
      "compression_loss": 0.0,
      "distillation_loss": 0.07206468284130096,
      "epoch": 4.05,
      "learning_rate": 1.892921004540507e-05,
      "loss": 0.0677,
      "step": 4264,
      "task_loss": 0.028736798092722893
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7879572708917829,
      "compression_loss": 0.0,
      "distillation_loss": 0.0855240449309349,
      "epoch": 4.05,
      "learning_rate": 1.8918874279299372e-05,
      "loss": 0.0786,
      "step": 4265,
      "task_loss": 0.01638454757630825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.787965953464706,
      "compression_loss": 0.0,
      "distillation_loss": 0.06442791223526001,
      "epoch": 4.05,
      "learning_rate": 1.8908539617859456e-05,
      "loss": 0.0701,
      "step": 4266,
      "task_loss": 0.12096739560365677
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7879746318633153,
      "compression_loss": 0.0,
      "distillation_loss": 0.07881754636764526,
      "epoch": 4.05,
      "learning_rate": 1.8898206062962647e-05,
      "loss": 0.0732,
      "step": 4267,
      "task_loss": 0.022822581231594086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7879833060886144,
      "compression_loss": 0.0,
      "distillation_loss": 0.09095026552677155,
      "epoch": 4.05,
      "learning_rate": 1.88878736164861e-05,
      "loss": 0.0839,
      "step": 4268,
      "task_loss": 0.020265545696020126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7879919761416071,
      "compression_loss": 0.0,
      "distillation_loss": 0.19944968819618225,
      "epoch": 4.05,
      "learning_rate": 1.8877542280306728e-05,
      "loss": 0.2095,
      "step": 4269,
      "task_loss": 0.2999950647354126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788000642023297,
      "compression_loss": 0.0,
      "distillation_loss": 0.08276577293872833,
      "epoch": 4.06,
      "learning_rate": 1.8867212056301305e-05,
      "loss": 0.0792,
      "step": 4270,
      "task_loss": 0.04671813175082207
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7880093037346878,
      "compression_loss": 0.0,
      "distillation_loss": 0.21997350454330444,
      "epoch": 4.06,
      "learning_rate": 1.8856882946346344e-05,
      "loss": 0.2087,
      "step": 4271,
      "task_loss": 0.10748877376317978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7880179612767833,
      "compression_loss": 0.0,
      "distillation_loss": 0.14946487545967102,
      "epoch": 4.06,
      "learning_rate": 1.8846554952318178e-05,
      "loss": 0.1507,
      "step": 4272,
      "task_loss": 0.1618417650461197
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7880266146505869,
      "compression_loss": 0.0,
      "distillation_loss": 0.25104832649230957,
      "epoch": 4.06,
      "learning_rate": 1.8836228076092945e-05,
      "loss": 0.2645,
      "step": 4273,
      "task_loss": 0.3852643370628357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7880352638571025,
      "compression_loss": 0.0,
      "distillation_loss": 0.01792140305042267,
      "epoch": 4.06,
      "learning_rate": 1.8825902319546565e-05,
      "loss": 0.0165,
      "step": 4274,
      "task_loss": 0.0033651478588581085
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7880439088973338,
      "compression_loss": 0.0,
      "distillation_loss": 0.051164254546165466,
      "epoch": 4.06,
      "learning_rate": 1.881557768455477e-05,
      "loss": 0.0474,
      "step": 4275,
      "task_loss": 0.013138506561517715
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7880525497722843,
      "compression_loss": 0.0,
      "distillation_loss": 0.1809595227241516,
      "epoch": 4.06,
      "learning_rate": 1.8805254172993064e-05,
      "loss": 0.1699,
      "step": 4276,
      "task_loss": 0.07031507790088654
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7880611864829579,
      "compression_loss": 0.0,
      "distillation_loss": 0.09073644876480103,
      "epoch": 4.06,
      "learning_rate": 1.879493178673677e-05,
      "loss": 0.0901,
      "step": 4277,
      "task_loss": 0.08396268635988235
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7880698190303581,
      "compression_loss": 0.0,
      "distillation_loss": 0.07403849810361862,
      "epoch": 4.06,
      "learning_rate": 1.8784610527661e-05,
      "loss": 0.0697,
      "step": 4278,
      "task_loss": 0.03097301721572876
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7880784474154886,
      "compression_loss": 0.0,
      "distillation_loss": 0.10361620783805847,
      "epoch": 4.06,
      "learning_rate": 1.8774290397640664e-05,
      "loss": 0.1067,
      "step": 4279,
      "task_loss": 0.13399073481559753
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7880870716393533,
      "compression_loss": 0.0,
      "distillation_loss": 0.04671156406402588,
      "epoch": 4.06,
      "learning_rate": 1.876397139855047e-05,
      "loss": 0.0516,
      "step": 4280,
      "task_loss": 0.09573487192392349
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7880956917029556,
      "compression_loss": 0.0,
      "distillation_loss": 0.06411126255989075,
      "epoch": 4.07,
      "learning_rate": 1.8753653532264894e-05,
      "loss": 0.0698,
      "step": 4281,
      "task_loss": 0.12079112231731415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7881043076072994,
      "compression_loss": 0.0,
      "distillation_loss": 0.10022042691707611,
      "epoch": 4.07,
      "learning_rate": 1.8743336800658245e-05,
      "loss": 0.1016,
      "step": 4282,
      "task_loss": 0.11449723690748215
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7881129193533882,
      "compression_loss": 0.0,
      "distillation_loss": 0.13063356280326843,
      "epoch": 4.07,
      "learning_rate": 1.8733021205604596e-05,
      "loss": 0.129,
      "step": 4283,
      "task_loss": 0.11380869895219803
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7881215269422258,
      "compression_loss": 0.0,
      "distillation_loss": 0.171952024102211,
      "epoch": 4.07,
      "learning_rate": 1.872270674897782e-05,
      "loss": 0.1669,
      "step": 4284,
      "task_loss": 0.12174376845359802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7881301303748159,
      "compression_loss": 0.0,
      "distillation_loss": 0.02115825191140175,
      "epoch": 4.07,
      "learning_rate": 1.8712393432651603e-05,
      "loss": 0.0195,
      "step": 4285,
      "task_loss": 0.0041760653257369995
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7881387296521621,
      "compression_loss": 0.0,
      "distillation_loss": 0.03141398727893829,
      "epoch": 4.07,
      "learning_rate": 1.87020812584994e-05,
      "loss": 0.0287,
      "step": 4286,
      "task_loss": 0.0046629346907138824
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7881473247752682,
      "compression_loss": 0.0,
      "distillation_loss": 0.10433746874332428,
      "epoch": 4.07,
      "learning_rate": 1.8691770228394456e-05,
      "loss": 0.1159,
      "step": 4287,
      "task_loss": 0.21988573670387268
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7881559157451377,
      "compression_loss": 0.0,
      "distillation_loss": 0.08133610337972641,
      "epoch": 4.07,
      "learning_rate": 1.868146034420984e-05,
      "loss": 0.0763,
      "step": 4288,
      "task_loss": 0.03072592243552208
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7881645025627744,
      "compression_loss": 0.0,
      "distillation_loss": 0.11363761126995087,
      "epoch": 4.07,
      "learning_rate": 1.8671151607818382e-05,
      "loss": 0.1087,
      "step": 4289,
      "task_loss": 0.06467755138874054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788173085229182,
      "compression_loss": 0.0,
      "distillation_loss": 0.051334887742996216,
      "epoch": 4.07,
      "learning_rate": 1.8660844021092716e-05,
      "loss": 0.0478,
      "step": 4290,
      "task_loss": 0.016054822131991386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788181663745364,
      "compression_loss": 0.0,
      "distillation_loss": 0.08099980652332306,
      "epoch": 4.08,
      "learning_rate": 1.8650537585905258e-05,
      "loss": 0.0847,
      "step": 4291,
      "task_loss": 0.11772287636995316
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7881902381123244,
      "compression_loss": 0.0,
      "distillation_loss": 0.015967125073075294,
      "epoch": 4.08,
      "learning_rate": 1.8640232304128236e-05,
      "loss": 0.0146,
      "step": 4292,
      "task_loss": 0.002349233254790306
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7881988083310666,
      "compression_loss": 0.0,
      "distillation_loss": 0.03628503531217575,
      "epoch": 4.08,
      "learning_rate": 1.8629928177633637e-05,
      "loss": 0.0488,
      "step": 4293,
      "task_loss": 0.16185718774795532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7882073744025945,
      "compression_loss": 0.0,
      "distillation_loss": 0.05933556705713272,
      "epoch": 4.08,
      "learning_rate": 1.8619625208293268e-05,
      "loss": 0.0568,
      "step": 4294,
      "task_loss": 0.03381485491991043
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7882159363279116,
      "compression_loss": 0.0,
      "distillation_loss": 0.16856390237808228,
      "epoch": 4.08,
      "learning_rate": 1.86093233979787e-05,
      "loss": 0.1643,
      "step": 4295,
      "task_loss": 0.12632951140403748
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7882244941080216,
      "compression_loss": 0.0,
      "distillation_loss": 0.07362917810678482,
      "epoch": 4.08,
      "learning_rate": 1.8599022748561325e-05,
      "loss": 0.075,
      "step": 4296,
      "task_loss": 0.08769040554761887
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7882330477439283,
      "compression_loss": 0.0,
      "distillation_loss": 0.03499433398246765,
      "epoch": 4.08,
      "learning_rate": 1.8588723261912288e-05,
      "loss": 0.0327,
      "step": 4297,
      "task_loss": 0.012281343340873718
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7882415972366354,
      "compression_loss": 0.0,
      "distillation_loss": 0.11919771134853363,
      "epoch": 4.08,
      "learning_rate": 1.857842493990255e-05,
      "loss": 0.1097,
      "step": 4298,
      "task_loss": 0.023989982903003693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7882501425871464,
      "compression_loss": 0.0,
      "distillation_loss": 0.11720117926597595,
      "epoch": 4.08,
      "learning_rate": 1.856812778440285e-05,
      "loss": 0.1113,
      "step": 4299,
      "task_loss": 0.057778194546699524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7882586837964651,
      "compression_loss": 0.0,
      "distillation_loss": 0.24679766595363617,
      "epoch": 4.08,
      "learning_rate": 1.8557831797283716e-05,
      "loss": 0.2454,
      "step": 4300,
      "task_loss": 0.23254123330116272
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7882672208655952,
      "compression_loss": 0.0,
      "distillation_loss": 0.05673684924840927,
      "epoch": 4.08,
      "learning_rate": 1.8547536980415452e-05,
      "loss": 0.0526,
      "step": 4301,
      "task_loss": 0.014921434223651886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7882757537955404,
      "compression_loss": 0.0,
      "distillation_loss": 0.06865353882312775,
      "epoch": 4.09,
      "learning_rate": 1.8537243335668187e-05,
      "loss": 0.0635,
      "step": 4302,
      "task_loss": 0.01695919781923294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7882842825873042,
      "compression_loss": 0.0,
      "distillation_loss": 0.08616451919078827,
      "epoch": 4.09,
      "learning_rate": 1.8526950864911784e-05,
      "loss": 0.0925,
      "step": 4303,
      "task_loss": 0.14940661191940308
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7882928072418904,
      "compression_loss": 0.0,
      "distillation_loss": 0.0653180330991745,
      "epoch": 4.09,
      "learning_rate": 1.8516659570015924e-05,
      "loss": 0.0677,
      "step": 4304,
      "task_loss": 0.08950569480657578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7883013277603028,
      "compression_loss": 0.0,
      "distillation_loss": 0.01897302269935608,
      "epoch": 4.09,
      "learning_rate": 1.8506369452850087e-05,
      "loss": 0.0177,
      "step": 4305,
      "task_loss": 0.0065006837248802185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788309844143545,
      "compression_loss": 0.0,
      "distillation_loss": 0.0725211426615715,
      "epoch": 4.09,
      "learning_rate": 1.8496080515283514e-05,
      "loss": 0.0802,
      "step": 4306,
      "task_loss": 0.14979489147663116
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7883183563926206,
      "compression_loss": 0.0,
      "distillation_loss": 0.03266327083110809,
      "epoch": 4.09,
      "learning_rate": 1.8485792759185232e-05,
      "loss": 0.0298,
      "step": 4307,
      "task_loss": 0.0037787500768899918
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7883268645085333,
      "compression_loss": 0.0,
      "distillation_loss": 0.02446456253528595,
      "epoch": 4.09,
      "learning_rate": 1.8475506186424074e-05,
      "loss": 0.0224,
      "step": 4308,
      "task_loss": 0.004015939310193062
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7883353684922869,
      "compression_loss": 0.0,
      "distillation_loss": 0.029062699526548386,
      "epoch": 4.09,
      "learning_rate": 1.846522079886864e-05,
      "loss": 0.0269,
      "step": 4309,
      "task_loss": 0.007744431495666504
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788343868344885,
      "compression_loss": 0.0,
      "distillation_loss": 0.026423152536153793,
      "epoch": 4.09,
      "learning_rate": 1.8454936598387317e-05,
      "loss": 0.0375,
      "step": 4310,
      "task_loss": 0.13689905405044556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7883523640673312,
      "compression_loss": 0.0,
      "distillation_loss": 0.18082647025585175,
      "epoch": 4.09,
      "learning_rate": 1.8444653586848286e-05,
      "loss": 0.1714,
      "step": 4311,
      "task_loss": 0.08704258501529694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7883608556606293,
      "compression_loss": 0.0,
      "distillation_loss": 0.06727072596549988,
      "epoch": 4.09,
      "learning_rate": 1.8434371766119496e-05,
      "loss": 0.0641,
      "step": 4312,
      "task_loss": 0.03516186401247978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788369343125783,
      "compression_loss": 0.0,
      "distillation_loss": 0.02708693966269493,
      "epoch": 4.1,
      "learning_rate": 1.8424091138068692e-05,
      "loss": 0.0254,
      "step": 4313,
      "task_loss": 0.009996037930250168
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788377826463796,
      "compression_loss": 0.0,
      "distillation_loss": 0.036099981516599655,
      "epoch": 4.1,
      "learning_rate": 1.8413811704563405e-05,
      "loss": 0.0384,
      "step": 4314,
      "task_loss": 0.059546999633312225
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7883863056756718,
      "compression_loss": 0.0,
      "distillation_loss": 0.048894353210926056,
      "epoch": 4.1,
      "learning_rate": 1.8403533467470946e-05,
      "loss": 0.0634,
      "step": 4315,
      "task_loss": 0.1943952441215515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7883947807624142,
      "compression_loss": 0.0,
      "distillation_loss": 0.1497052013874054,
      "epoch": 4.1,
      "learning_rate": 1.8393256428658403e-05,
      "loss": 0.1447,
      "step": 4316,
      "task_loss": 0.0995352566242218
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7884032517250269,
      "compression_loss": 0.0,
      "distillation_loss": 0.07365826517343521,
      "epoch": 4.1,
      "learning_rate": 1.8382980589992643e-05,
      "loss": 0.0695,
      "step": 4317,
      "task_loss": 0.031793296337127686
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7884117185645135,
      "compression_loss": 0.0,
      "distillation_loss": 0.058896757662296295,
      "epoch": 4.1,
      "learning_rate": 1.8372705953340337e-05,
      "loss": 0.0537,
      "step": 4318,
      "task_loss": 0.006742917001247406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7884201812818779,
      "compression_loss": 0.0,
      "distillation_loss": 0.057655736804008484,
      "epoch": 4.1,
      "learning_rate": 1.8362432520567903e-05,
      "loss": 0.0562,
      "step": 4319,
      "task_loss": 0.04272696375846863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7884286398781235,
      "compression_loss": 0.0,
      "distillation_loss": 0.06040318310260773,
      "epoch": 4.1,
      "learning_rate": 1.8352160293541566e-05,
      "loss": 0.0561,
      "step": 4320,
      "task_loss": 0.017850998789072037
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7884370943542541,
      "compression_loss": 0.0,
      "distillation_loss": 0.028085466474294662,
      "epoch": 4.1,
      "learning_rate": 1.834188927412732e-05,
      "loss": 0.0255,
      "step": 4321,
      "task_loss": 0.002726750448346138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7884455447112735,
      "compression_loss": 0.0,
      "distillation_loss": 0.014654599130153656,
      "epoch": 4.1,
      "learning_rate": 1.833161946419097e-05,
      "loss": 0.0203,
      "step": 4322,
      "task_loss": 0.07066480070352554
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7884539909501852,
      "compression_loss": 0.0,
      "distillation_loss": 0.14747686684131622,
      "epoch": 4.11,
      "learning_rate": 1.8321350865598057e-05,
      "loss": 0.1489,
      "step": 4323,
      "task_loss": 0.16213619709014893
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7884624330719929,
      "compression_loss": 0.0,
      "distillation_loss": 0.030887076631188393,
      "epoch": 4.11,
      "learning_rate": 1.831108348021392e-05,
      "loss": 0.0283,
      "step": 4324,
      "task_loss": 0.005010116845369339
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7884708710777004,
      "compression_loss": 0.0,
      "distillation_loss": 0.056699033826589584,
      "epoch": 4.11,
      "learning_rate": 1.8300817309903686e-05,
      "loss": 0.0685,
      "step": 4325,
      "task_loss": 0.17428170144557953
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7884793049683114,
      "compression_loss": 0.0,
      "distillation_loss": 0.03099614754319191,
      "epoch": 4.11,
      "learning_rate": 1.8290552356532247e-05,
      "loss": 0.0287,
      "step": 4326,
      "task_loss": 0.008489016443490982
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7884877347448294,
      "compression_loss": 0.0,
      "distillation_loss": 0.04078204557299614,
      "epoch": 4.11,
      "learning_rate": 1.8280288621964288e-05,
      "loss": 0.0456,
      "step": 4327,
      "task_loss": 0.08923979848623276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7884961604082582,
      "compression_loss": 0.0,
      "distillation_loss": 0.0556565560400486,
      "epoch": 4.11,
      "learning_rate": 1.827002610806427e-05,
      "loss": 0.0541,
      "step": 4328,
      "task_loss": 0.03984564542770386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7885045819596015,
      "compression_loss": 0.0,
      "distillation_loss": 0.06724396347999573,
      "epoch": 4.11,
      "learning_rate": 1.825976481669641e-05,
      "loss": 0.0626,
      "step": 4329,
      "task_loss": 0.02046894282102585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7885129993998629,
      "compression_loss": 0.0,
      "distillation_loss": 0.034551121294498444,
      "epoch": 4.11,
      "learning_rate": 1.824950474972473e-05,
      "loss": 0.0469,
      "step": 4330,
      "task_loss": 0.15822812914848328
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7885214127300461,
      "compression_loss": 0.0,
      "distillation_loss": 0.03939036652445793,
      "epoch": 4.11,
      "learning_rate": 1.823924590901303e-05,
      "loss": 0.0436,
      "step": 4331,
      "task_loss": 0.08171267807483673
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7885298219511548,
      "compression_loss": 0.0,
      "distillation_loss": 0.046313732862472534,
      "epoch": 4.11,
      "learning_rate": 1.8228988296424877e-05,
      "loss": 0.0426,
      "step": 4332,
      "task_loss": 0.009343873709440231
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7885382270641927,
      "compression_loss": 0.0,
      "distillation_loss": 0.16361156105995178,
      "epoch": 4.11,
      "learning_rate": 1.82187319138236e-05,
      "loss": 0.1613,
      "step": 4333,
      "task_loss": 0.14066770672798157
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7885466280701636,
      "compression_loss": 0.0,
      "distillation_loss": 0.12138652801513672,
      "epoch": 4.12,
      "learning_rate": 1.8208476763072332e-05,
      "loss": 0.1152,
      "step": 4334,
      "task_loss": 0.05983586981892586
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788555024970071,
      "compression_loss": 0.0,
      "distillation_loss": 0.05892729386687279,
      "epoch": 4.12,
      "learning_rate": 1.8198222846033975e-05,
      "loss": 0.0612,
      "step": 4335,
      "task_loss": 0.0814908966422081
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7885634177649186,
      "compression_loss": 0.0,
      "distillation_loss": 0.01657470129430294,
      "epoch": 4.12,
      "learning_rate": 1.8187970164571187e-05,
      "loss": 0.0158,
      "step": 4336,
      "task_loss": 0.008594617247581482
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7885718064557101,
      "compression_loss": 0.0,
      "distillation_loss": 0.16905143857002258,
      "epoch": 4.12,
      "learning_rate": 1.817771872054643e-05,
      "loss": 0.16,
      "step": 4337,
      "task_loss": 0.07822901755571365
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7885801910434493,
      "compression_loss": 0.0,
      "distillation_loss": 0.12039444595575333,
      "epoch": 4.12,
      "learning_rate": 1.8167468515821924e-05,
      "loss": 0.1294,
      "step": 4338,
      "task_loss": 0.210740864276886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7885885715291397,
      "compression_loss": 0.0,
      "distillation_loss": 0.320880651473999,
      "epoch": 4.12,
      "learning_rate": 1.815721955225966e-05,
      "loss": 0.3213,
      "step": 4339,
      "task_loss": 0.32472920417785645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7885969479137851,
      "compression_loss": 0.0,
      "distillation_loss": 0.11701447516679764,
      "epoch": 4.12,
      "learning_rate": 1.8146971831721426e-05,
      "loss": 0.1076,
      "step": 4340,
      "task_loss": 0.023150721564888954
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7886053201983891,
      "compression_loss": 0.0,
      "distillation_loss": 0.05050637945532799,
      "epoch": 4.12,
      "learning_rate": 1.8136725356068762e-05,
      "loss": 0.0468,
      "step": 4341,
      "task_loss": 0.013836957514286041
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7886136883839555,
      "compression_loss": 0.0,
      "distillation_loss": 0.14656773209571838,
      "epoch": 4.12,
      "learning_rate": 1.8126480127163e-05,
      "loss": 0.1385,
      "step": 4342,
      "task_loss": 0.06583775579929352
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788622052471488,
      "compression_loss": 0.0,
      "distillation_loss": 0.17784056067466736,
      "epoch": 4.12,
      "learning_rate": 1.8116236146865213e-05,
      "loss": 0.1849,
      "step": 4343,
      "task_loss": 0.24860072135925293
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7886304124619901,
      "compression_loss": 0.0,
      "distillation_loss": 0.03818412870168686,
      "epoch": 4.13,
      "learning_rate": 1.810599341703629e-05,
      "loss": 0.0394,
      "step": 4344,
      "task_loss": 0.0502682588994503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7886387683564655,
      "compression_loss": 0.0,
      "distillation_loss": 0.1047634482383728,
      "epoch": 4.13,
      "learning_rate": 1.8095751939536866e-05,
      "loss": 0.0986,
      "step": 4345,
      "task_loss": 0.04341891035437584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788647120155918,
      "compression_loss": 0.0,
      "distillation_loss": 0.26263827085494995,
      "epoch": 4.13,
      "learning_rate": 1.8085511716227345e-05,
      "loss": 0.255,
      "step": 4346,
      "task_loss": 0.18654467165470123
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7886554678613513,
      "compression_loss": 0.0,
      "distillation_loss": 0.09224887192249298,
      "epoch": 4.13,
      "learning_rate": 1.807527274896792e-05,
      "loss": 0.0892,
      "step": 4347,
      "task_loss": 0.06142377480864525
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788663811473769,
      "compression_loss": 0.0,
      "distillation_loss": 0.12528349459171295,
      "epoch": 4.13,
      "learning_rate": 1.8065035039618556e-05,
      "loss": 0.1222,
      "step": 4348,
      "task_loss": 0.09476065635681152
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7886721509941748,
      "compression_loss": 0.0,
      "distillation_loss": 0.04307865723967552,
      "epoch": 4.13,
      "learning_rate": 1.8054798590038984e-05,
      "loss": 0.0394,
      "step": 4349,
      "task_loss": 0.006099509075284004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7886804864235724,
      "compression_loss": 0.0,
      "distillation_loss": 0.046839505434036255,
      "epoch": 4.13,
      "learning_rate": 1.8044563402088684e-05,
      "loss": 0.0433,
      "step": 4350,
      "task_loss": 0.01109330728650093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7886888177629654,
      "compression_loss": 0.0,
      "distillation_loss": 0.03697451949119568,
      "epoch": 4.13,
      "learning_rate": 1.8034329477626945e-05,
      "loss": 0.0493,
      "step": 4351,
      "task_loss": 0.16054922342300415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7886971450133576,
      "compression_loss": 0.0,
      "distillation_loss": 0.066026471555233,
      "epoch": 4.13,
      "learning_rate": 1.8024096818512807e-05,
      "loss": 0.0672,
      "step": 4352,
      "task_loss": 0.07799084484577179
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7887054681757527,
      "compression_loss": 0.0,
      "distillation_loss": 0.047900401055812836,
      "epoch": 4.13,
      "learning_rate": 1.8013865426605076e-05,
      "loss": 0.0578,
      "step": 4353,
      "task_loss": 0.14682576060295105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7887137872511543,
      "compression_loss": 0.0,
      "distillation_loss": 0.04553601145744324,
      "epoch": 4.13,
      "learning_rate": 1.8003635303762336e-05,
      "loss": 0.0483,
      "step": 4354,
      "task_loss": 0.0731407031416893
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788722102240566,
      "compression_loss": 0.0,
      "distillation_loss": 0.07538627833127975,
      "epoch": 4.14,
      "learning_rate": 1.7993406451842935e-05,
      "loss": 0.0774,
      "step": 4355,
      "task_loss": 0.0952514261007309
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7887304131449917,
      "compression_loss": 0.0,
      "distillation_loss": 0.2628774642944336,
      "epoch": 4.14,
      "learning_rate": 1.7983178872704992e-05,
      "loss": 0.2549,
      "step": 4356,
      "task_loss": 0.1826838105916977
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7887387199654349,
      "compression_loss": 0.0,
      "distillation_loss": 0.10139012336730957,
      "epoch": 4.14,
      "learning_rate": 1.7972952568206402e-05,
      "loss": 0.1041,
      "step": 4357,
      "task_loss": 0.12895506620407104
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7887470227028993,
      "compression_loss": 0.0,
      "distillation_loss": 0.19154085218906403,
      "epoch": 4.14,
      "learning_rate": 1.7962727540204827e-05,
      "loss": 0.1773,
      "step": 4358,
      "task_loss": 0.04891242831945419
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7887553213583888,
      "compression_loss": 0.0,
      "distillation_loss": 0.09380476176738739,
      "epoch": 4.14,
      "learning_rate": 1.7952503790557686e-05,
      "loss": 0.097,
      "step": 4359,
      "task_loss": 0.1261102557182312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7887636159329067,
      "compression_loss": 0.0,
      "distillation_loss": 0.10539199411869049,
      "epoch": 4.14,
      "learning_rate": 1.7942281321122168e-05,
      "loss": 0.1145,
      "step": 4360,
      "task_loss": 0.19654574990272522
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788771906427457,
      "compression_loss": 0.0,
      "distillation_loss": 0.021082423627376556,
      "epoch": 4.14,
      "learning_rate": 1.7932060133755245e-05,
      "loss": 0.0256,
      "step": 4361,
      "task_loss": 0.06635300070047379
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7887801928430432,
      "compression_loss": 0.0,
      "distillation_loss": 0.020916704088449478,
      "epoch": 4.14,
      "learning_rate": 1.792184023031363e-05,
      "loss": 0.0241,
      "step": 4362,
      "task_loss": 0.05260460451245308
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7887884751806692,
      "compression_loss": 0.0,
      "distillation_loss": 0.13232557475566864,
      "epoch": 4.14,
      "learning_rate": 1.7911621612653832e-05,
      "loss": 0.1251,
      "step": 4363,
      "task_loss": 0.060476090759038925
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7887967534413384,
      "compression_loss": 0.0,
      "distillation_loss": 0.1687774807214737,
      "epoch": 4.14,
      "learning_rate": 1.7901404282632105e-05,
      "loss": 0.156,
      "step": 4364,
      "task_loss": 0.04108697548508644
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7888050276260546,
      "compression_loss": 0.0,
      "distillation_loss": 0.052245475351810455,
      "epoch": 4.15,
      "learning_rate": 1.7891188242104466e-05,
      "loss": 0.0564,
      "step": 4365,
      "task_loss": 0.09406599402427673
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7888132977358217,
      "compression_loss": 0.0,
      "distillation_loss": 0.02484312653541565,
      "epoch": 4.15,
      "learning_rate": 1.7880973492926734e-05,
      "loss": 0.0229,
      "step": 4366,
      "task_loss": 0.005430508404970169
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788821563771643,
      "compression_loss": 0.0,
      "distillation_loss": 0.06918057799339294,
      "epoch": 4.15,
      "learning_rate": 1.7870760036954444e-05,
      "loss": 0.0648,
      "step": 4367,
      "task_loss": 0.02556915022432804
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7888298257345224,
      "compression_loss": 0.0,
      "distillation_loss": 0.19346660375595093,
      "epoch": 4.15,
      "learning_rate": 1.786054787604294e-05,
      "loss": 0.1922,
      "step": 4368,
      "task_loss": 0.18102970719337463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7888380836254636,
      "compression_loss": 0.0,
      "distillation_loss": 0.1463518738746643,
      "epoch": 4.15,
      "learning_rate": 1.7850337012047287e-05,
      "loss": 0.155,
      "step": 4369,
      "task_loss": 0.23283621668815613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7888463374454703,
      "compression_loss": 0.0,
      "distillation_loss": 0.06679415702819824,
      "epoch": 4.15,
      "learning_rate": 1.784012744682235e-05,
      "loss": 0.0708,
      "step": 4370,
      "task_loss": 0.10708191245794296
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7888545871955459,
      "compression_loss": 0.0,
      "distillation_loss": 0.015153437852859497,
      "epoch": 4.15,
      "learning_rate": 1.7829919182222752e-05,
      "loss": 0.0192,
      "step": 4371,
      "task_loss": 0.0555480532348156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7888628328766945,
      "compression_loss": 0.0,
      "distillation_loss": 0.04128699749708176,
      "epoch": 4.15,
      "learning_rate": 1.7819712220102857e-05,
      "loss": 0.0446,
      "step": 4372,
      "task_loss": 0.07435785233974457
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7888710744899194,
      "compression_loss": 0.0,
      "distillation_loss": 0.07817722856998444,
      "epoch": 4.15,
      "learning_rate": 1.7809506562316818e-05,
      "loss": 0.08,
      "step": 4373,
      "task_loss": 0.09686526656150818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7888793120362246,
      "compression_loss": 0.0,
      "distillation_loss": 0.058174848556518555,
      "epoch": 4.15,
      "learning_rate": 1.7799302210718544e-05,
      "loss": 0.0739,
      "step": 4374,
      "task_loss": 0.21509107947349548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7888875455166136,
      "compression_loss": 0.0,
      "distillation_loss": 0.02214631997048855,
      "epoch": 4.15,
      "learning_rate": 1.7789099167161704e-05,
      "loss": 0.0367,
      "step": 4375,
      "task_loss": 0.1679784506559372
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7888957749320901,
      "compression_loss": 0.0,
      "distillation_loss": 0.03224210441112518,
      "epoch": 4.16,
      "learning_rate": 1.777889743349973e-05,
      "loss": 0.0327,
      "step": 4376,
      "task_loss": 0.036665864288806915
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7889040002836578,
      "compression_loss": 0.0,
      "distillation_loss": 0.04349253326654434,
      "epoch": 4.16,
      "learning_rate": 1.776869701158581e-05,
      "loss": 0.041,
      "step": 4377,
      "task_loss": 0.01853703148663044
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7889122215723204,
      "compression_loss": 0.0,
      "distillation_loss": 0.03350648283958435,
      "epoch": 4.16,
      "learning_rate": 1.775849790327291e-05,
      "loss": 0.0312,
      "step": 4378,
      "task_loss": 0.010053610429167747
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7889204387990816,
      "compression_loss": 0.0,
      "distillation_loss": 0.11184386909008026,
      "epoch": 4.16,
      "learning_rate": 1.7748300110413737e-05,
      "loss": 0.105,
      "step": 4379,
      "task_loss": 0.04385565221309662
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.788928651964945,
      "compression_loss": 0.0,
      "distillation_loss": 0.022888878360390663,
      "epoch": 4.16,
      "learning_rate": 1.7738103634860776e-05,
      "loss": 0.0224,
      "step": 4380,
      "task_loss": 0.017647787928581238
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7889368610709144,
      "compression_loss": 0.0,
      "distillation_loss": 0.11289818584918976,
      "epoch": 4.16,
      "learning_rate": 1.7727908478466264e-05,
      "loss": 0.1062,
      "step": 4381,
      "task_loss": 0.04638146981596947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7889450661179934,
      "compression_loss": 0.0,
      "distillation_loss": 0.1492418795824051,
      "epoch": 4.16,
      "learning_rate": 1.771771464308219e-05,
      "loss": 0.1528,
      "step": 4382,
      "task_loss": 0.18500253558158875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7889532671071856,
      "compression_loss": 0.0,
      "distillation_loss": 0.13683563470840454,
      "epoch": 4.16,
      "learning_rate": 1.770752213056033e-05,
      "loss": 0.137,
      "step": 4383,
      "task_loss": 0.13819673657417297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7889614640394949,
      "compression_loss": 0.0,
      "distillation_loss": 0.02154296077787876,
      "epoch": 4.16,
      "learning_rate": 1.7697330942752193e-05,
      "loss": 0.0206,
      "step": 4384,
      "task_loss": 0.012236261740326881
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7889696569159248,
      "compression_loss": 0.0,
      "distillation_loss": 0.041471950709819794,
      "epoch": 4.16,
      "learning_rate": 1.768714108150907e-05,
      "loss": 0.044,
      "step": 4385,
      "task_loss": 0.06637803465127945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7889778457374791,
      "compression_loss": 0.0,
      "distillation_loss": 0.1052827313542366,
      "epoch": 4.17,
      "learning_rate": 1.767695254868198e-05,
      "loss": 0.1006,
      "step": 4386,
      "task_loss": 0.057975780218839645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7889860305051614,
      "compression_loss": 0.0,
      "distillation_loss": 0.0629846528172493,
      "epoch": 4.17,
      "learning_rate": 1.766676534612173e-05,
      "loss": 0.0609,
      "step": 4387,
      "task_loss": 0.042560774832963943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7889942112199754,
      "compression_loss": 0.0,
      "distillation_loss": 0.039940863847732544,
      "epoch": 4.17,
      "learning_rate": 1.7656579475678876e-05,
      "loss": 0.0388,
      "step": 4388,
      "task_loss": 0.028355613350868225
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7890023878829248,
      "compression_loss": 0.0,
      "distillation_loss": 0.03699345886707306,
      "epoch": 4.17,
      "learning_rate": 1.764639493920372e-05,
      "loss": 0.041,
      "step": 4389,
      "task_loss": 0.07748576998710632
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7890105604950132,
      "compression_loss": 0.0,
      "distillation_loss": 0.03237525001168251,
      "epoch": 4.17,
      "learning_rate": 1.763621173854635e-05,
      "loss": 0.0299,
      "step": 4390,
      "task_loss": 0.007551593706011772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7890187290572445,
      "compression_loss": 0.0,
      "distillation_loss": 0.08855956047773361,
      "epoch": 4.17,
      "learning_rate": 1.762602987555656e-05,
      "loss": 0.0878,
      "step": 4391,
      "task_loss": 0.08059325814247131
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7890268935706221,
      "compression_loss": 0.0,
      "distillation_loss": 0.04030374437570572,
      "epoch": 4.17,
      "learning_rate": 1.7615849352083975e-05,
      "loss": 0.0561,
      "step": 4392,
      "task_loss": 0.1978634148836136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7890350540361498,
      "compression_loss": 0.0,
      "distillation_loss": 0.021272670477628708,
      "epoch": 4.17,
      "learning_rate": 1.760567016997791e-05,
      "loss": 0.0197,
      "step": 4393,
      "task_loss": 0.005058445036411285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7890432104548314,
      "compression_loss": 0.0,
      "distillation_loss": 0.13198691606521606,
      "epoch": 4.17,
      "learning_rate": 1.760567016997791e-05,
      "loss": 0.128,
      "step": 4394,
      "task_loss": 0.09192854911088943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7890513628276704,
      "compression_loss": 0.0,
      "distillation_loss": 0.14407461881637573,
      "epoch": 4.17,
      "learning_rate": 1.7595492331087472e-05,
      "loss": 0.1359,
      "step": 4395,
      "task_loss": 0.06224524974822998
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7890595111556706,
      "compression_loss": 0.0,
      "distillation_loss": 0.028794020414352417,
      "epoch": 4.17,
      "learning_rate": 1.7585315837261518e-05,
      "loss": 0.0571,
      "step": 4396,
      "task_loss": 0.31160321831703186
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7890676554398357,
      "compression_loss": 0.0,
      "distillation_loss": 0.1819879114627838,
      "epoch": 4.18,
      "learning_rate": 1.7575140690348647e-05,
      "loss": 0.1749,
      "step": 4397,
      "task_loss": 0.11067160218954086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7890757956811694,
      "compression_loss": 0.0,
      "distillation_loss": 0.0241214781999588,
      "epoch": 4.18,
      "learning_rate": 1.756496689219723e-05,
      "loss": 0.0231,
      "step": 4398,
      "task_loss": 0.013442834839224815
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7890839318806752,
      "compression_loss": 0.0,
      "distillation_loss": 0.0341639518737793,
      "epoch": 4.18,
      "learning_rate": 1.755479444465538e-05,
      "loss": 0.0382,
      "step": 4399,
      "task_loss": 0.07429169118404388
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7890920640393568,
      "compression_loss": 0.0,
      "distillation_loss": 0.06670385599136353,
      "epoch": 4.18,
      "learning_rate": 1.7544623349570973e-05,
      "loss": 0.071,
      "step": 4400,
      "task_loss": 0.11016245931386948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7891001921582181,
      "compression_loss": 0.0,
      "distillation_loss": 0.11859071999788284,
      "epoch": 4.18,
      "learning_rate": 1.7534453608791644e-05,
      "loss": 0.1156,
      "step": 4401,
      "task_loss": 0.08839291334152222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7891083162382626,
      "compression_loss": 0.0,
      "distillation_loss": 0.027910087257623672,
      "epoch": 4.18,
      "learning_rate": 1.7524285224164772e-05,
      "loss": 0.0256,
      "step": 4402,
      "task_loss": 0.005291949957609177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7891164362804941,
      "compression_loss": 0.0,
      "distillation_loss": 0.05852048099040985,
      "epoch": 4.18,
      "learning_rate": 1.7514118197537497e-05,
      "loss": 0.0757,
      "step": 4403,
      "task_loss": 0.2301456481218338
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7891245522859162,
      "compression_loss": 0.0,
      "distillation_loss": 0.0825999528169632,
      "epoch": 4.18,
      "learning_rate": 1.75039525307567e-05,
      "loss": 0.091,
      "step": 4404,
      "task_loss": 0.1666475236415863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7891326642555326,
      "compression_loss": 0.0,
      "distillation_loss": 0.02013680338859558,
      "epoch": 4.18,
      "learning_rate": 1.7493788225669027e-05,
      "loss": 0.0268,
      "step": 4405,
      "task_loss": 0.08626913279294968
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789140772190347,
      "compression_loss": 0.0,
      "distillation_loss": 0.024234000593423843,
      "epoch": 4.18,
      "learning_rate": 1.7483625284120876e-05,
      "loss": 0.031,
      "step": 4406,
      "task_loss": 0.09184201061725616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7891488760913631,
      "compression_loss": 0.0,
      "distillation_loss": 0.11344337463378906,
      "epoch": 4.19,
      "learning_rate": 1.7473463707958388e-05,
      "loss": 0.1076,
      "step": 4407,
      "task_loss": 0.055413682013750076
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7891569759595845,
      "compression_loss": 0.0,
      "distillation_loss": 0.021596118807792664,
      "epoch": 4.19,
      "learning_rate": 1.7463303499027466e-05,
      "loss": 0.0199,
      "step": 4408,
      "task_loss": 0.004350002855062485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7891650717960149,
      "compression_loss": 0.0,
      "distillation_loss": 0.03837398812174797,
      "epoch": 4.19,
      "learning_rate": 1.745314465917375e-05,
      "loss": 0.0356,
      "step": 4409,
      "task_loss": 0.010302269831299782
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789173163601658,
      "compression_loss": 0.0,
      "distillation_loss": 0.23183999955654144,
      "epoch": 4.19,
      "learning_rate": 1.7442987190242668e-05,
      "loss": 0.2387,
      "step": 4410,
      "task_loss": 0.3002890646457672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7891812513775176,
      "compression_loss": 0.0,
      "distillation_loss": 0.1880335658788681,
      "epoch": 4.19,
      "learning_rate": 1.7432831094079355e-05,
      "loss": 0.1926,
      "step": 4411,
      "task_loss": 0.2336317002773285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7891893351245972,
      "compression_loss": 0.0,
      "distillation_loss": 0.13378804922103882,
      "epoch": 4.19,
      "learning_rate": 1.7422676372528718e-05,
      "loss": 0.13,
      "step": 4412,
      "task_loss": 0.09611691534519196
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7891974148439006,
      "compression_loss": 0.0,
      "distillation_loss": 0.021651534363627434,
      "epoch": 4.19,
      "learning_rate": 1.7412523027435407e-05,
      "loss": 0.0242,
      "step": 4413,
      "task_loss": 0.04686147719621658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7892054905364314,
      "compression_loss": 0.0,
      "distillation_loss": 0.10245324671268463,
      "epoch": 4.19,
      "learning_rate": 1.740237106064383e-05,
      "loss": 0.1,
      "step": 4414,
      "task_loss": 0.07831554114818573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7892135622031934,
      "compression_loss": 0.0,
      "distillation_loss": 0.016463253647089005,
      "epoch": 4.19,
      "learning_rate": 1.7392220473998147e-05,
      "loss": 0.0154,
      "step": 4415,
      "task_loss": 0.005847932770848274
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7892216298451902,
      "compression_loss": 0.0,
      "distillation_loss": 0.050443872809410095,
      "epoch": 4.19,
      "learning_rate": 1.738207126934225e-05,
      "loss": 0.0581,
      "step": 4416,
      "task_loss": 0.12747488915920258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7892296934634254,
      "compression_loss": 0.0,
      "distillation_loss": 0.11745072156190872,
      "epoch": 4.19,
      "learning_rate": 1.737192344851979e-05,
      "loss": 0.112,
      "step": 4417,
      "task_loss": 0.06290709227323532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7892377530589028,
      "compression_loss": 0.0,
      "distillation_loss": 0.041940174996852875,
      "epoch": 4.2,
      "learning_rate": 1.7361777013374173e-05,
      "loss": 0.0392,
      "step": 4418,
      "task_loss": 0.01443542167544365
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7892458086326261,
      "compression_loss": 0.0,
      "distillation_loss": 0.18550318479537964,
      "epoch": 4.2,
      "learning_rate": 1.7351631965748555e-05,
      "loss": 0.1769,
      "step": 4419,
      "task_loss": 0.09899017959833145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7892538601855988,
      "compression_loss": 0.0,
      "distillation_loss": 0.02498285286128521,
      "epoch": 4.2,
      "learning_rate": 1.734148830748582e-05,
      "loss": 0.023,
      "step": 4420,
      "task_loss": 0.0047489944845438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7892619077188249,
      "compression_loss": 0.0,
      "distillation_loss": 0.02473044954240322,
      "epoch": 4.2,
      "learning_rate": 1.733134604042862e-05,
      "loss": 0.0284,
      "step": 4421,
      "task_loss": 0.06133115291595459
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7892699512333078,
      "compression_loss": 0.0,
      "distillation_loss": 0.07032322883605957,
      "epoch": 4.2,
      "learning_rate": 1.7321205166419348e-05,
      "loss": 0.0686,
      "step": 4422,
      "task_loss": 0.052621208131313324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7892779907300513,
      "compression_loss": 0.0,
      "distillation_loss": 0.06657904386520386,
      "epoch": 4.2,
      "learning_rate": 1.7311065687300133e-05,
      "loss": 0.0622,
      "step": 4423,
      "task_loss": 0.022893797606229782
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789286026210059,
      "compression_loss": 0.0,
      "distillation_loss": 0.06309516727924347,
      "epoch": 4.2,
      "learning_rate": 1.730092760491287e-05,
      "loss": 0.0599,
      "step": 4424,
      "task_loss": 0.031243909150362015
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7892940576743348,
      "compression_loss": 0.0,
      "distillation_loss": 0.07031765580177307,
      "epoch": 4.2,
      "learning_rate": 1.729079092109919e-05,
      "loss": 0.0728,
      "step": 4425,
      "task_loss": 0.09468785673379898
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7893020851238821,
      "compression_loss": 0.0,
      "distillation_loss": 0.2251608967781067,
      "epoch": 4.2,
      "learning_rate": 1.7280655637700456e-05,
      "loss": 0.222,
      "step": 4426,
      "task_loss": 0.1932929903268814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7893101085597048,
      "compression_loss": 0.0,
      "distillation_loss": 0.022628076374530792,
      "epoch": 4.2,
      "learning_rate": 1.7270521756557805e-05,
      "loss": 0.0233,
      "step": 4427,
      "task_loss": 0.02976515144109726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7893181279828064,
      "compression_loss": 0.0,
      "distillation_loss": 0.040559072047472,
      "epoch": 4.21,
      "learning_rate": 1.7260389279512106e-05,
      "loss": 0.0407,
      "step": 4428,
      "task_loss": 0.041948944330215454
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7893261433941907,
      "compression_loss": 0.0,
      "distillation_loss": 0.02462448924779892,
      "epoch": 4.21,
      "learning_rate": 1.7250258208403974e-05,
      "loss": 0.0236,
      "step": 4429,
      "task_loss": 0.014600781723856926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7893341547948615,
      "compression_loss": 0.0,
      "distillation_loss": 0.02943100593984127,
      "epoch": 4.21,
      "learning_rate": 1.7240128545073753e-05,
      "loss": 0.0367,
      "step": 4430,
      "task_loss": 0.10226895660161972
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7893421621858222,
      "compression_loss": 0.0,
      "distillation_loss": 0.23634879291057587,
      "epoch": 4.21,
      "learning_rate": 1.723000029136156e-05,
      "loss": 0.244,
      "step": 4431,
      "task_loss": 0.3131251037120819
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7893501655680766,
      "compression_loss": 0.0,
      "distillation_loss": 0.14374813437461853,
      "epoch": 4.21,
      "learning_rate": 1.7219873449107233e-05,
      "loss": 0.136,
      "step": 4432,
      "task_loss": 0.06602862477302551
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7893581649426284,
      "compression_loss": 0.0,
      "distillation_loss": 0.03242877870798111,
      "epoch": 4.21,
      "learning_rate": 1.7209748020150362e-05,
      "loss": 0.0366,
      "step": 4433,
      "task_loss": 0.07460720837116241
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7893661603104813,
      "compression_loss": 0.0,
      "distillation_loss": 0.028279315680265427,
      "epoch": 4.21,
      "learning_rate": 1.719962400633028e-05,
      "loss": 0.0334,
      "step": 4434,
      "task_loss": 0.07948870211839676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789374151672639,
      "compression_loss": 0.0,
      "distillation_loss": 0.021993128582835197,
      "epoch": 4.21,
      "learning_rate": 1.7189501409486062e-05,
      "loss": 0.0263,
      "step": 4435,
      "task_loss": 0.06503792107105255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7893821390301051,
      "compression_loss": 0.0,
      "distillation_loss": 0.08905947208404541,
      "epoch": 4.21,
      "learning_rate": 1.717938023145654e-05,
      "loss": 0.103,
      "step": 4436,
      "task_loss": 0.22882232069969177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7893901223838834,
      "compression_loss": 0.0,
      "distillation_loss": 0.02391093783080578,
      "epoch": 4.21,
      "learning_rate": 1.716926047408025e-05,
      "loss": 0.022,
      "step": 4437,
      "task_loss": 0.004962872713804245
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7893981017349775,
      "compression_loss": 0.0,
      "distillation_loss": 0.1042497307062149,
      "epoch": 4.21,
      "learning_rate": 1.7159142139195514e-05,
      "loss": 0.1155,
      "step": 4438,
      "task_loss": 0.21651120483875275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7894060770843911,
      "compression_loss": 0.0,
      "distillation_loss": 0.14275360107421875,
      "epoch": 4.22,
      "learning_rate": 1.7149025228640376e-05,
      "loss": 0.1396,
      "step": 4439,
      "task_loss": 0.11126483976840973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7894140484331279,
      "compression_loss": 0.0,
      "distillation_loss": 0.0388154536485672,
      "epoch": 4.22,
      "learning_rate": 1.7138909744252608e-05,
      "loss": 0.0382,
      "step": 4440,
      "task_loss": 0.03281474858522415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7894220157821915,
      "compression_loss": 0.0,
      "distillation_loss": 0.07178475707769394,
      "epoch": 4.22,
      "learning_rate": 1.712879568786975e-05,
      "loss": 0.0693,
      "step": 4441,
      "task_loss": 0.046773068606853485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7894299791325856,
      "compression_loss": 0.0,
      "distillation_loss": 0.02249164879322052,
      "epoch": 4.22,
      "learning_rate": 1.711868306132906e-05,
      "loss": 0.0268,
      "step": 4442,
      "task_loss": 0.06524014472961426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789437938485314,
      "compression_loss": 0.0,
      "distillation_loss": 0.028957027941942215,
      "epoch": 4.22,
      "learning_rate": 1.7108571866467547e-05,
      "loss": 0.0456,
      "step": 4443,
      "task_loss": 0.19559374451637268
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7894458938413803,
      "compression_loss": 0.0,
      "distillation_loss": 0.13259674608707428,
      "epoch": 4.22,
      "learning_rate": 1.709846210512196e-05,
      "loss": 0.1353,
      "step": 4444,
      "task_loss": 0.15936444699764252
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7894538452017882,
      "compression_loss": 0.0,
      "distillation_loss": 0.18647846579551697,
      "epoch": 4.22,
      "learning_rate": 1.7088353779128784e-05,
      "loss": 0.1902,
      "step": 4445,
      "task_loss": 0.22320255637168884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7894617925675413,
      "compression_loss": 0.0,
      "distillation_loss": 0.12089197337627411,
      "epoch": 4.22,
      "learning_rate": 1.7078246890324257e-05,
      "loss": 0.1152,
      "step": 4446,
      "task_loss": 0.06432140618562698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7894697359396434,
      "compression_loss": 0.0,
      "distillation_loss": 0.016367292031645775,
      "epoch": 4.22,
      "learning_rate": 1.706814144054433e-05,
      "loss": 0.0152,
      "step": 4447,
      "task_loss": 0.004441702738404274
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7894776753190981,
      "compression_loss": 0.0,
      "distillation_loss": 0.02612762153148651,
      "epoch": 4.22,
      "learning_rate": 1.705803743162471e-05,
      "loss": 0.0238,
      "step": 4448,
      "task_loss": 0.0026707202196121216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7894856107069091,
      "compression_loss": 0.0,
      "distillation_loss": 0.046527355909347534,
      "epoch": 4.23,
      "learning_rate": 1.704793486540084e-05,
      "loss": 0.0495,
      "step": 4449,
      "task_loss": 0.07587733864784241
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7894935421040802,
      "compression_loss": 0.0,
      "distillation_loss": 0.016864405944943428,
      "epoch": 4.23,
      "learning_rate": 1.7037833743707892e-05,
      "loss": 0.0237,
      "step": 4450,
      "task_loss": 0.08481664955615997
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7895014695116149,
      "compression_loss": 0.0,
      "distillation_loss": 0.026964813470840454,
      "epoch": 4.23,
      "learning_rate": 1.7027734068380803e-05,
      "loss": 0.0318,
      "step": 4451,
      "task_loss": 0.07501386851072311
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789509392930517,
      "compression_loss": 0.0,
      "distillation_loss": 0.0899241715669632,
      "epoch": 4.23,
      "learning_rate": 1.7017635841254194e-05,
      "loss": 0.0956,
      "step": 4452,
      "task_loss": 0.14717119932174683
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7895173123617901,
      "compression_loss": 0.0,
      "distillation_loss": 0.07356803119182587,
      "epoch": 4.23,
      "learning_rate": 1.7007539064162498e-05,
      "loss": 0.0707,
      "step": 4453,
      "task_loss": 0.04454321041703224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789525227806438,
      "compression_loss": 0.0,
      "distillation_loss": 0.09831836819648743,
      "epoch": 4.23,
      "learning_rate": 1.6997443738939815e-05,
      "loss": 0.0933,
      "step": 4454,
      "task_loss": 0.04812869057059288
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7895331392654642,
      "compression_loss": 0.0,
      "distillation_loss": 0.036545686423778534,
      "epoch": 4.23,
      "learning_rate": 1.6987349867420024e-05,
      "loss": 0.0339,
      "step": 4455,
      "task_loss": 0.010073903948068619
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7895410467398726,
      "compression_loss": 0.0,
      "distillation_loss": 0.15401822328567505,
      "epoch": 4.23,
      "learning_rate": 1.6977257451436712e-05,
      "loss": 0.1472,
      "step": 4456,
      "task_loss": 0.08590954542160034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7895489502306667,
      "compression_loss": 0.0,
      "distillation_loss": 0.14784236252307892,
      "epoch": 4.23,
      "learning_rate": 1.6967166492823226e-05,
      "loss": 0.1489,
      "step": 4457,
      "task_loss": 0.15834221243858337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7895568497388503,
      "compression_loss": 0.0,
      "distillation_loss": 0.09763270616531372,
      "epoch": 4.23,
      "learning_rate": 1.6957076993412636e-05,
      "loss": 0.0944,
      "step": 4458,
      "task_loss": 0.06483699381351471
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789564745265427,
      "compression_loss": 0.0,
      "distillation_loss": 0.03840739279985428,
      "epoch": 4.23,
      "learning_rate": 1.694698895503774e-05,
      "loss": 0.0356,
      "step": 4459,
      "task_loss": 0.010582242161035538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7895726368114006,
      "compression_loss": 0.0,
      "distillation_loss": 0.04345633089542389,
      "epoch": 4.24,
      "learning_rate": 1.6936902379531082e-05,
      "loss": 0.0488,
      "step": 4460,
      "task_loss": 0.09703336656093597
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7895805243777746,
      "compression_loss": 0.0,
      "distillation_loss": 0.0363914854824543,
      "epoch": 4.24,
      "learning_rate": 1.6926817268724938e-05,
      "loss": 0.0399,
      "step": 4461,
      "task_loss": 0.07188586890697479
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7895884079655529,
      "compression_loss": 0.0,
      "distillation_loss": 0.14720529317855835,
      "epoch": 4.24,
      "learning_rate": 1.6916733624451324e-05,
      "loss": 0.1541,
      "step": 4462,
      "task_loss": 0.2162376046180725
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7895962875757389,
      "compression_loss": 0.0,
      "distillation_loss": 0.07019872963428497,
      "epoch": 4.24,
      "learning_rate": 1.690665144854198e-05,
      "loss": 0.0753,
      "step": 4463,
      "task_loss": 0.12100522965192795
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7896041632093366,
      "compression_loss": 0.0,
      "distillation_loss": 0.03042607009410858,
      "epoch": 4.24,
      "learning_rate": 1.6896570742828367e-05,
      "loss": 0.0286,
      "step": 4464,
      "task_loss": 0.012003440409898758
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7896120348673494,
      "compression_loss": 0.0,
      "distillation_loss": 0.10173506289720535,
      "epoch": 4.24,
      "learning_rate": 1.6886491509141717e-05,
      "loss": 0.1217,
      "step": 4465,
      "task_loss": 0.3017996847629547
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7896199025507812,
      "compression_loss": 0.0,
      "distillation_loss": 0.04094619303941727,
      "epoch": 4.24,
      "learning_rate": 1.6876413749312954e-05,
      "loss": 0.0476,
      "step": 4466,
      "task_loss": 0.10766998678445816
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7896277662606356,
      "compression_loss": 0.0,
      "distillation_loss": 0.019710058346390724,
      "epoch": 4.24,
      "learning_rate": 1.6866337465172754e-05,
      "loss": 0.0182,
      "step": 4467,
      "task_loss": 0.004470134153962135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7896356259979161,
      "compression_loss": 0.0,
      "distillation_loss": 0.04553668200969696,
      "epoch": 4.24,
      "learning_rate": 1.685626265855153e-05,
      "loss": 0.0429,
      "step": 4468,
      "task_loss": 0.01903415471315384
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7896434817636268,
      "compression_loss": 0.0,
      "distillation_loss": 0.03563941270112991,
      "epoch": 4.24,
      "learning_rate": 1.6846189331279415e-05,
      "loss": 0.0426,
      "step": 4469,
      "task_loss": 0.10529813170433044
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789651333558771,
      "compression_loss": 0.0,
      "distillation_loss": 0.09041808545589447,
      "epoch": 4.25,
      "learning_rate": 1.683611748518627e-05,
      "loss": 0.0915,
      "step": 4470,
      "task_loss": 0.10150431841611862
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7896591813843525,
      "compression_loss": 0.0,
      "distillation_loss": 0.031974878162145615,
      "epoch": 4.25,
      "learning_rate": 1.6826047122101703e-05,
      "loss": 0.0362,
      "step": 4471,
      "task_loss": 0.07398272305727005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7896670252413751,
      "compression_loss": 0.0,
      "distillation_loss": 0.03553183376789093,
      "epoch": 4.25,
      "learning_rate": 1.6815978243855052e-05,
      "loss": 0.0332,
      "step": 4472,
      "task_loss": 0.012660248205065727
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7896748651308424,
      "compression_loss": 0.0,
      "distillation_loss": 0.1044113039970398,
      "epoch": 4.25,
      "learning_rate": 1.6805910852275358e-05,
      "loss": 0.1011,
      "step": 4473,
      "task_loss": 0.07151272147893906
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789682701053758,
      "compression_loss": 0.0,
      "distillation_loss": 0.1547268033027649,
      "epoch": 4.25,
      "learning_rate": 1.6795844949191426e-05,
      "loss": 0.1734,
      "step": 4474,
      "task_loss": 0.34136468172073364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7896905330111257,
      "compression_loss": 0.0,
      "distillation_loss": 0.07073305547237396,
      "epoch": 4.25,
      "learning_rate": 1.6785780536431772e-05,
      "loss": 0.0803,
      "step": 4475,
      "task_loss": 0.16646496951580048
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7896983610039491,
      "compression_loss": 0.0,
      "distillation_loss": 0.0397295281291008,
      "epoch": 4.25,
      "learning_rate": 1.677571761582464e-05,
      "loss": 0.0411,
      "step": 4476,
      "task_loss": 0.05347587913274765
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789706185033232,
      "compression_loss": 0.0,
      "distillation_loss": 0.04183419793844223,
      "epoch": 4.25,
      "learning_rate": 1.6765656189198013e-05,
      "loss": 0.0536,
      "step": 4477,
      "task_loss": 0.1599850058555603
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7897140050999778,
      "compression_loss": 0.0,
      "distillation_loss": 0.10868053883314133,
      "epoch": 4.25,
      "learning_rate": 1.675559625837959e-05,
      "loss": 0.1083,
      "step": 4478,
      "task_loss": 0.10476875305175781
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7897218212051905,
      "compression_loss": 0.0,
      "distillation_loss": 0.15036195516586304,
      "epoch": 4.25,
      "learning_rate": 1.6745537825196823e-05,
      "loss": 0.144,
      "step": 4479,
      "task_loss": 0.08634737133979797
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7897296333498737,
      "compression_loss": 0.0,
      "distillation_loss": 0.12207113206386566,
      "epoch": 4.25,
      "learning_rate": 1.6735480891476855e-05,
      "loss": 0.1236,
      "step": 4480,
      "task_loss": 0.13720379769802094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789737441535031,
      "compression_loss": 0.0,
      "distillation_loss": 0.030313938856124878,
      "epoch": 4.26,
      "learning_rate": 1.672542545904659e-05,
      "loss": 0.0284,
      "step": 4481,
      "task_loss": 0.010765250772237778
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7897452457616662,
      "compression_loss": 0.0,
      "distillation_loss": 0.08106144517660141,
      "epoch": 4.26,
      "learning_rate": 1.6715371529732643e-05,
      "loss": 0.1023,
      "step": 4482,
      "task_loss": 0.2930178940296173
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7897530460307828,
      "compression_loss": 0.0,
      "distillation_loss": 0.21419498324394226,
      "epoch": 4.26,
      "learning_rate": 1.6705319105361357e-05,
      "loss": 0.2036,
      "step": 4483,
      "task_loss": 0.10846757143735886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7897608423433846,
      "compression_loss": 0.0,
      "distillation_loss": 0.09862993657588959,
      "epoch": 4.26,
      "learning_rate": 1.6695268187758797e-05,
      "loss": 0.1024,
      "step": 4484,
      "task_loss": 0.1367725133895874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7897686347004753,
      "compression_loss": 0.0,
      "distillation_loss": 0.05723030865192413,
      "epoch": 4.26,
      "learning_rate": 1.6685218778750775e-05,
      "loss": 0.0594,
      "step": 4485,
      "task_loss": 0.07874364405870438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7897764231030585,
      "compression_loss": 0.0,
      "distillation_loss": 0.017719998955726624,
      "epoch": 4.26,
      "learning_rate": 1.66751708801628e-05,
      "loss": 0.0165,
      "step": 4486,
      "task_loss": 0.005893906578421593
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789784207552138,
      "compression_loss": 0.0,
      "distillation_loss": 0.06374501436948776,
      "epoch": 4.26,
      "learning_rate": 1.6665124493820123e-05,
      "loss": 0.0772,
      "step": 4487,
      "task_loss": 0.19814634323120117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7897919880487174,
      "compression_loss": 0.0,
      "distillation_loss": 0.05392298847436905,
      "epoch": 4.26,
      "learning_rate": 1.6655079621547727e-05,
      "loss": 0.0499,
      "step": 4488,
      "task_loss": 0.013967299833893776
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7897997645938004,
      "compression_loss": 0.0,
      "distillation_loss": 0.05942397564649582,
      "epoch": 4.26,
      "learning_rate": 1.6645036265170314e-05,
      "loss": 0.0579,
      "step": 4489,
      "task_loss": 0.04388638213276863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7898075371883907,
      "compression_loss": 0.0,
      "distillation_loss": 0.035995546728372574,
      "epoch": 4.26,
      "learning_rate": 1.6634994426512296e-05,
      "loss": 0.0414,
      "step": 4490,
      "task_loss": 0.09044382721185684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7898153058334919,
      "compression_loss": 0.0,
      "distillation_loss": 0.07521656155586243,
      "epoch": 4.26,
      "learning_rate": 1.662495410739783e-05,
      "loss": 0.0794,
      "step": 4491,
      "task_loss": 0.11709287762641907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7898230705301078,
      "compression_loss": 0.0,
      "distillation_loss": 0.04527838155627251,
      "epoch": 4.27,
      "learning_rate": 1.661491530965078e-05,
      "loss": 0.0513,
      "step": 4492,
      "task_loss": 0.10566423833370209
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789830831279242,
      "compression_loss": 0.0,
      "distillation_loss": 0.017616283148527145,
      "epoch": 4.27,
      "learning_rate": 1.660487803509475e-05,
      "loss": 0.0164,
      "step": 4493,
      "task_loss": 0.0058811865746974945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7898385880818982,
      "compression_loss": 0.0,
      "distillation_loss": 0.04209098592400551,
      "epoch": 4.27,
      "learning_rate": 1.6594842285553062e-05,
      "loss": 0.0466,
      "step": 4494,
      "task_loss": 0.08670932054519653
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7898463409390801,
      "compression_loss": 0.0,
      "distillation_loss": 0.022504784166812897,
      "epoch": 4.27,
      "learning_rate": 1.6584808062848743e-05,
      "loss": 0.0301,
      "step": 4495,
      "task_loss": 0.09832281619310379
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7898540898517915,
      "compression_loss": 0.0,
      "distillation_loss": 0.09364865720272064,
      "epoch": 4.27,
      "learning_rate": 1.6574775368804567e-05,
      "loss": 0.0985,
      "step": 4496,
      "task_loss": 0.14257903397083282
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7898618348210358,
      "compression_loss": 0.0,
      "distillation_loss": 0.03564376011490822,
      "epoch": 4.27,
      "learning_rate": 1.656474420524302e-05,
      "loss": 0.0466,
      "step": 4497,
      "task_loss": 0.14488337934017181
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7898695758478169,
      "compression_loss": 0.0,
      "distillation_loss": 0.0824960246682167,
      "epoch": 4.27,
      "learning_rate": 1.6554714573986324e-05,
      "loss": 0.0832,
      "step": 4498,
      "task_loss": 0.08933614194393158
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7898773129331385,
      "compression_loss": 0.0,
      "distillation_loss": 0.14729326963424683,
      "epoch": 4.27,
      "learning_rate": 1.654468647685639e-05,
      "loss": 0.1449,
      "step": 4499,
      "task_loss": 0.12342742830514908
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.789885046078004,
      "compression_loss": 0.0,
      "distillation_loss": 0.02807687409222126,
      "epoch": 4.27,
      "learning_rate": 1.6534659915674882e-05,
      "loss": 0.0259,
      "step": 4500,
      "task_loss": 0.006013935431838036
    },
    {
      "epoch": 4.27,
      "eval_accuracy": 0.8979357798165137,
      "eval_loss": 0.482150673866272,
      "eval_runtime": 18.0094,
      "eval_samples_per_second": 48.419,
      "eval_steps_per_second": 6.052,
      "step": 4500
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7898927752834174,
      "compression_loss": 0.0,
      "distillation_loss": 0.05030520260334015,
      "epoch": 4.27,
      "learning_rate": 1.6524634892263176e-05,
      "loss": 0.049,
      "step": 4501,
      "task_loss": 0.0376378558576107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7899005005503823,
      "compression_loss": 0.0,
      "distillation_loss": 0.10964888334274292,
      "epoch": 4.28,
      "learning_rate": 1.651461140844235e-05,
      "loss": 0.1056,
      "step": 4502,
      "task_loss": 0.0688520222902298
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7899082218799023,
      "compression_loss": 0.0,
      "distillation_loss": 0.19415023922920227,
      "epoch": 4.28,
      "learning_rate": 1.6504589466033226e-05,
      "loss": 0.1881,
      "step": 4503,
      "task_loss": 0.133355051279068
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7899159392729812,
      "compression_loss": 0.0,
      "distillation_loss": 0.021605759859085083,
      "epoch": 4.28,
      "learning_rate": 1.6494569066856343e-05,
      "loss": 0.0294,
      "step": 4504,
      "task_loss": 0.09938529133796692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7899236527306225,
      "compression_loss": 0.0,
      "distillation_loss": 0.025142306461930275,
      "epoch": 4.28,
      "learning_rate": 1.6484550212731953e-05,
      "loss": 0.0293,
      "step": 4505,
      "task_loss": 0.06672917306423187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7899313622538301,
      "compression_loss": 0.0,
      "distillation_loss": 0.07510918378829956,
      "epoch": 4.28,
      "learning_rate": 1.6474532905480027e-05,
      "loss": 0.0898,
      "step": 4506,
      "task_loss": 0.22238320112228394
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7899390678436075,
      "compression_loss": 0.0,
      "distillation_loss": 0.06910199671983719,
      "epoch": 4.28,
      "learning_rate": 1.6464517146920255e-05,
      "loss": 0.0727,
      "step": 4507,
      "task_loss": 0.10474320501089096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7899467695009585,
      "compression_loss": 0.0,
      "distillation_loss": 0.023270685225725174,
      "epoch": 4.28,
      "learning_rate": 1.645450293887206e-05,
      "loss": 0.0214,
      "step": 4508,
      "task_loss": 0.004897216334939003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7899544672268868,
      "compression_loss": 0.0,
      "distillation_loss": 0.01756512001156807,
      "epoch": 4.28,
      "learning_rate": 1.6444490283154557e-05,
      "loss": 0.0164,
      "step": 4509,
      "task_loss": 0.006135221570730209
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7899621610223959,
      "compression_loss": 0.0,
      "distillation_loss": 0.05131693556904793,
      "epoch": 4.28,
      "learning_rate": 1.6434479181586594e-05,
      "loss": 0.067,
      "step": 4510,
      "task_loss": 0.20782402157783508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7899698508884897,
      "compression_loss": 0.0,
      "distillation_loss": 0.024594193324446678,
      "epoch": 4.28,
      "learning_rate": 1.6424469635986744e-05,
      "loss": 0.0271,
      "step": 4511,
      "task_loss": 0.04964712634682655
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7899775368261717,
      "compression_loss": 0.0,
      "distillation_loss": 0.0458107516169548,
      "epoch": 4.28,
      "learning_rate": 1.6414461648173284e-05,
      "loss": 0.0431,
      "step": 4512,
      "task_loss": 0.019116627052426338
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7899852188364457,
      "compression_loss": 0.0,
      "distillation_loss": 0.028601201251149178,
      "epoch": 4.29,
      "learning_rate": 1.6404455219964203e-05,
      "loss": 0.0261,
      "step": 4513,
      "task_loss": 0.0036617927253246307
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7899928969203154,
      "compression_loss": 0.0,
      "distillation_loss": 0.07332905381917953,
      "epoch": 4.29,
      "learning_rate": 1.6394450353177242e-05,
      "loss": 0.0782,
      "step": 4514,
      "task_loss": 0.12209247052669525
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7900005710787844,
      "compression_loss": 0.0,
      "distillation_loss": 0.051219768822193146,
      "epoch": 4.29,
      "learning_rate": 1.6384447049629816e-05,
      "loss": 0.0545,
      "step": 4515,
      "task_loss": 0.08445730060338974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7900082413128564,
      "compression_loss": 0.0,
      "distillation_loss": 0.12552213668823242,
      "epoch": 4.29,
      "learning_rate": 1.6374445311139074e-05,
      "loss": 0.1197,
      "step": 4516,
      "task_loss": 0.06750532984733582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7900159076235351,
      "compression_loss": 0.0,
      "distillation_loss": 0.231977179646492,
      "epoch": 4.29,
      "learning_rate": 1.6364445139521883e-05,
      "loss": 0.2244,
      "step": 4517,
      "task_loss": 0.15585389733314514
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7900235700118242,
      "compression_loss": 0.0,
      "distillation_loss": 0.027736982330679893,
      "epoch": 4.29,
      "learning_rate": 1.635444653659483e-05,
      "loss": 0.0257,
      "step": 4518,
      "task_loss": 0.007768923416733742
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7900312284787274,
      "compression_loss": 0.0,
      "distillation_loss": 0.050665318965911865,
      "epoch": 4.29,
      "learning_rate": 1.6344449504174193e-05,
      "loss": 0.0625,
      "step": 4519,
      "task_loss": 0.16943272948265076
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7900388830252483,
      "compression_loss": 0.0,
      "distillation_loss": 0.03559410944581032,
      "epoch": 4.29,
      "learning_rate": 1.6334454044075988e-05,
      "loss": 0.0405,
      "step": 4520,
      "task_loss": 0.08508558571338654
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7900465336523906,
      "compression_loss": 0.0,
      "distillation_loss": 0.05828773230314255,
      "epoch": 4.29,
      "learning_rate": 1.6324460158115942e-05,
      "loss": 0.0552,
      "step": 4521,
      "task_loss": 0.027413969859480858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7900541803611582,
      "compression_loss": 0.0,
      "distillation_loss": 0.2908591628074646,
      "epoch": 4.29,
      "learning_rate": 1.6314467848109483e-05,
      "loss": 0.2765,
      "step": 4522,
      "task_loss": 0.14718219637870789
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7900618231525544,
      "compression_loss": 0.0,
      "distillation_loss": 0.013387423008680344,
      "epoch": 4.3,
      "learning_rate": 1.6304477115871776e-05,
      "loss": 0.0194,
      "step": 4523,
      "task_loss": 0.07351858913898468
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7900694620275831,
      "compression_loss": 0.0,
      "distillation_loss": 0.13720417022705078,
      "epoch": 4.3,
      "learning_rate": 1.6294487963217677e-05,
      "loss": 0.129,
      "step": 4524,
      "task_loss": 0.05525549128651619
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7900770969872479,
      "compression_loss": 0.0,
      "distillation_loss": 0.024422302842140198,
      "epoch": 4.3,
      "learning_rate": 1.6284500391961772e-05,
      "loss": 0.0328,
      "step": 4525,
      "task_loss": 0.10812153667211533
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7900847280325526,
      "compression_loss": 0.0,
      "distillation_loss": 0.13294796645641327,
      "epoch": 4.3,
      "learning_rate": 1.627451440391834e-05,
      "loss": 0.1583,
      "step": 4526,
      "task_loss": 0.38683199882507324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7900923551645008,
      "compression_loss": 0.0,
      "distillation_loss": 0.025384480133652687,
      "epoch": 4.3,
      "learning_rate": 1.626453000090139e-05,
      "loss": 0.0332,
      "step": 4527,
      "task_loss": 0.10327724367380142
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7900999783840963,
      "compression_loss": 0.0,
      "distillation_loss": 0.1557433307170868,
      "epoch": 4.3,
      "learning_rate": 1.625454718472464e-05,
      "loss": 0.1661,
      "step": 4528,
      "task_loss": 0.2594253718852997
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7901075976923425,
      "compression_loss": 0.0,
      "distillation_loss": 0.07799336314201355,
      "epoch": 4.3,
      "learning_rate": 1.6244565957201506e-05,
      "loss": 0.0748,
      "step": 4529,
      "task_loss": 0.04625723510980606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7901152130902435,
      "compression_loss": 0.0,
      "distillation_loss": 0.03408731520175934,
      "epoch": 4.3,
      "learning_rate": 1.6234586320145125e-05,
      "loss": 0.0314,
      "step": 4530,
      "task_loss": 0.007508426904678345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7901228245788026,
      "compression_loss": 0.0,
      "distillation_loss": 0.023436367511749268,
      "epoch": 4.3,
      "learning_rate": 1.6224608275368364e-05,
      "loss": 0.0238,
      "step": 4531,
      "task_loss": 0.026913581416010857
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7901304321590236,
      "compression_loss": 0.0,
      "distillation_loss": 0.02828259766101837,
      "epoch": 4.3,
      "learning_rate": 1.6214631824683773e-05,
      "loss": 0.026,
      "step": 4532,
      "task_loss": 0.005199538543820381
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7901380358319102,
      "compression_loss": 0.0,
      "distillation_loss": 0.02083125337958336,
      "epoch": 4.3,
      "learning_rate": 1.6204656969903618e-05,
      "loss": 0.0344,
      "step": 4533,
      "task_loss": 0.15678700804710388
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7901456355984662,
      "compression_loss": 0.0,
      "distillation_loss": 0.023461364209651947,
      "epoch": 4.31,
      "learning_rate": 1.6194683712839885e-05,
      "loss": 0.0259,
      "step": 4534,
      "task_loss": 0.048007965087890625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7901532314596952,
      "compression_loss": 0.0,
      "distillation_loss": 0.03079340234398842,
      "epoch": 4.31,
      "learning_rate": 1.618471205530427e-05,
      "loss": 0.0345,
      "step": 4535,
      "task_loss": 0.0679367259144783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7901608234166008,
      "compression_loss": 0.0,
      "distillation_loss": 0.034932032227516174,
      "epoch": 4.31,
      "learning_rate": 1.6174741999108157e-05,
      "loss": 0.0328,
      "step": 4536,
      "task_loss": 0.013840943574905396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7901684114701867,
      "compression_loss": 0.0,
      "distillation_loss": 0.09038272500038147,
      "epoch": 4.31,
      "learning_rate": 1.6164773546062667e-05,
      "loss": 0.0854,
      "step": 4537,
      "task_loss": 0.04105079919099808
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7901759956214566,
      "compression_loss": 0.0,
      "distillation_loss": 0.056141406297683716,
      "epoch": 4.31,
      "learning_rate": 1.6154806697978608e-05,
      "loss": 0.0534,
      "step": 4538,
      "task_loss": 0.028462648391723633
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7901835758714143,
      "compression_loss": 0.0,
      "distillation_loss": 0.02680385112762451,
      "epoch": 4.31,
      "learning_rate": 1.614484145666651e-05,
      "loss": 0.0247,
      "step": 4539,
      "task_loss": 0.005452977493405342
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7901911522210633,
      "compression_loss": 0.0,
      "distillation_loss": 0.036854278296232224,
      "epoch": 4.31,
      "learning_rate": 1.613487782393661e-05,
      "loss": 0.0338,
      "step": 4540,
      "task_loss": 0.006693465635180473
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7901987246714074,
      "compression_loss": 0.0,
      "distillation_loss": 0.019134201109409332,
      "epoch": 4.31,
      "learning_rate": 1.6124915801598852e-05,
      "loss": 0.0225,
      "step": 4541,
      "task_loss": 0.05329040437936783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7902062932234503,
      "compression_loss": 0.0,
      "distillation_loss": 0.2091977298259735,
      "epoch": 4.31,
      "learning_rate": 1.6114955391462878e-05,
      "loss": 0.2147,
      "step": 4542,
      "task_loss": 0.2644974887371063
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7902138578781955,
      "compression_loss": 0.0,
      "distillation_loss": 0.07729081809520721,
      "epoch": 4.31,
      "learning_rate": 1.6104996595338047e-05,
      "loss": 0.0728,
      "step": 4543,
      "task_loss": 0.03260170668363571
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.790221418636647,
      "compression_loss": 0.0,
      "distillation_loss": 0.1044703871011734,
      "epoch": 4.32,
      "learning_rate": 1.609503941503343e-05,
      "loss": 0.107,
      "step": 4544,
      "task_loss": 0.12950289249420166
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7902289754998082,
      "compression_loss": 0.0,
      "distillation_loss": 0.03879034146666527,
      "epoch": 4.32,
      "learning_rate": 1.6085083852357786e-05,
      "loss": 0.0542,
      "step": 4545,
      "task_loss": 0.1928063929080963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7902365284686829,
      "compression_loss": 0.0,
      "distillation_loss": 0.13962599635124207,
      "epoch": 4.32,
      "learning_rate": 1.6075129909119592e-05,
      "loss": 0.1528,
      "step": 4546,
      "task_loss": 0.27165764570236206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7902440775442747,
      "compression_loss": 0.0,
      "distillation_loss": 0.05956469848752022,
      "epoch": 4.32,
      "learning_rate": 1.6065177587127027e-05,
      "loss": 0.0687,
      "step": 4547,
      "task_loss": 0.15051259100437164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7902516227275874,
      "compression_loss": 0.0,
      "distillation_loss": 0.022979095578193665,
      "epoch": 4.32,
      "learning_rate": 1.6055226888188e-05,
      "loss": 0.0238,
      "step": 4548,
      "task_loss": 0.030920779332518578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7902591640196246,
      "compression_loss": 0.0,
      "distillation_loss": 0.015515048988163471,
      "epoch": 4.32,
      "learning_rate": 1.604527781411008e-05,
      "loss": 0.0215,
      "step": 4549,
      "task_loss": 0.07502803206443787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79026670142139,
      "compression_loss": 0.0,
      "distillation_loss": 0.025966979563236237,
      "epoch": 4.32,
      "learning_rate": 1.6035330366700567e-05,
      "loss": 0.039,
      "step": 4550,
      "task_loss": 0.15588192641735077
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7902742349338874,
      "compression_loss": 0.0,
      "distillation_loss": 0.011777522042393684,
      "epoch": 4.32,
      "learning_rate": 1.6025384547766477e-05,
      "loss": 0.011,
      "step": 4551,
      "task_loss": 0.003892483189702034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7902817645581203,
      "compression_loss": 0.0,
      "distillation_loss": 0.018658041954040527,
      "epoch": 4.32,
      "learning_rate": 1.6015440359114497e-05,
      "loss": 0.0173,
      "step": 4552,
      "task_loss": 0.004857182502746582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7902892902950923,
      "compression_loss": 0.0,
      "distillation_loss": 0.052327413111925125,
      "epoch": 4.32,
      "learning_rate": 1.600549780255105e-05,
      "loss": 0.0637,
      "step": 4553,
      "task_loss": 0.1657162308692932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7902968121458074,
      "compression_loss": 0.0,
      "distillation_loss": 0.22867000102996826,
      "epoch": 4.32,
      "learning_rate": 1.5995556879882246e-05,
      "loss": 0.2293,
      "step": 4554,
      "task_loss": 0.23492863774299622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7903043301112691,
      "compression_loss": 0.0,
      "distillation_loss": 0.014244308695197105,
      "epoch": 4.33,
      "learning_rate": 1.59856175929139e-05,
      "loss": 0.0134,
      "step": 4555,
      "task_loss": 0.006260443478822708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7903118441924811,
      "compression_loss": 0.0,
      "distillation_loss": 0.06026815250515938,
      "epoch": 4.33,
      "learning_rate": 1.597567994345152e-05,
      "loss": 0.0547,
      "step": 4556,
      "task_loss": 0.004877146333456039
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7903193543904471,
      "compression_loss": 0.0,
      "distillation_loss": 0.04478321969509125,
      "epoch": 4.33,
      "learning_rate": 1.5965743933300352e-05,
      "loss": 0.0435,
      "step": 4557,
      "task_loss": 0.03176284581422806
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7903268607061708,
      "compression_loss": 0.0,
      "distillation_loss": 0.044730305671691895,
      "epoch": 4.33,
      "learning_rate": 1.595580956426531e-05,
      "loss": 0.0555,
      "step": 4558,
      "task_loss": 0.1524236649274826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7903343631406556,
      "compression_loss": 0.0,
      "distillation_loss": 0.022832248359918594,
      "epoch": 4.33,
      "learning_rate": 1.5945876838151014e-05,
      "loss": 0.0285,
      "step": 4559,
      "task_loss": 0.07918474823236465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7903418616949057,
      "compression_loss": 0.0,
      "distillation_loss": 0.13835708796977997,
      "epoch": 4.33,
      "learning_rate": 1.5935945756761794e-05,
      "loss": 0.1415,
      "step": 4560,
      "task_loss": 0.17012692987918854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7903493563699244,
      "compression_loss": 0.0,
      "distillation_loss": 0.06880520284175873,
      "epoch": 4.33,
      "learning_rate": 1.592601632190169e-05,
      "loss": 0.0639,
      "step": 4561,
      "task_loss": 0.019452493637800217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7903568471667155,
      "compression_loss": 0.0,
      "distillation_loss": 0.046133361756801605,
      "epoch": 4.33,
      "learning_rate": 1.591608853537441e-05,
      "loss": 0.0463,
      "step": 4562,
      "task_loss": 0.04781344532966614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7903643340862827,
      "compression_loss": 0.0,
      "distillation_loss": 0.038969479501247406,
      "epoch": 4.33,
      "learning_rate": 1.5906162398983397e-05,
      "loss": 0.05,
      "step": 4563,
      "task_loss": 0.14975658059120178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7903718171296297,
      "compression_loss": 0.0,
      "distillation_loss": 0.07891889661550522,
      "epoch": 4.33,
      "learning_rate": 1.5896237914531793e-05,
      "loss": 0.0827,
      "step": 4564,
      "task_loss": 0.1163320243358612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79037929629776,
      "compression_loss": 0.0,
      "distillation_loss": 0.014497898519039154,
      "epoch": 4.34,
      "learning_rate": 1.58863150838224e-05,
      "loss": 0.0135,
      "step": 4565,
      "task_loss": 0.004507582634687424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7903867715916776,
      "compression_loss": 0.0,
      "distillation_loss": 0.07853427529335022,
      "epoch": 4.34,
      "learning_rate": 1.5876393908657766e-05,
      "loss": 0.0769,
      "step": 4566,
      "task_loss": 0.06254800409078598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7903942430123858,
      "compression_loss": 0.0,
      "distillation_loss": 0.04548817127943039,
      "epoch": 4.34,
      "learning_rate": 1.5866474390840125e-05,
      "loss": 0.0488,
      "step": 4567,
      "task_loss": 0.07903194427490234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7904017105608887,
      "compression_loss": 0.0,
      "distillation_loss": 0.01196481566876173,
      "epoch": 4.34,
      "learning_rate": 1.58565565321714e-05,
      "loss": 0.0192,
      "step": 4568,
      "task_loss": 0.0844261422753334
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7904091742381896,
      "compression_loss": 0.0,
      "distillation_loss": 0.0883462056517601,
      "epoch": 4.34,
      "learning_rate": 1.584664033445321e-05,
      "loss": 0.1073,
      "step": 4569,
      "task_loss": 0.27824994921684265
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7904166340452924,
      "compression_loss": 0.0,
      "distillation_loss": 0.06949096918106079,
      "epoch": 4.34,
      "learning_rate": 1.583672579948689e-05,
      "loss": 0.0762,
      "step": 4570,
      "task_loss": 0.13672670722007751
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7904240899832008,
      "compression_loss": 0.0,
      "distillation_loss": 0.017165692523121834,
      "epoch": 4.34,
      "learning_rate": 1.582681292907346e-05,
      "loss": 0.0189,
      "step": 4571,
      "task_loss": 0.03450107201933861
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7904315420529183,
      "compression_loss": 0.0,
      "distillation_loss": 0.028937380760908127,
      "epoch": 4.34,
      "learning_rate": 1.581690172501364e-05,
      "loss": 0.0334,
      "step": 4572,
      "task_loss": 0.07330554723739624
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7904389902554487,
      "compression_loss": 0.0,
      "distillation_loss": 0.04542282223701477,
      "epoch": 4.34,
      "learning_rate": 1.5806992189107838e-05,
      "loss": 0.0445,
      "step": 4573,
      "task_loss": 0.036514170467853546
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7904464345917958,
      "compression_loss": 0.0,
      "distillation_loss": 0.11212094128131866,
      "epoch": 4.34,
      "learning_rate": 1.5797084323156186e-05,
      "loss": 0.1122,
      "step": 4574,
      "task_loss": 0.11315297335386276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.790453875062963,
      "compression_loss": 0.0,
      "distillation_loss": 0.09616538882255554,
      "epoch": 4.34,
      "learning_rate": 1.5787178128958496e-05,
      "loss": 0.0895,
      "step": 4575,
      "task_loss": 0.029983239248394966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7904613116699543,
      "compression_loss": 0.0,
      "distillation_loss": 0.03785378485918045,
      "epoch": 4.35,
      "learning_rate": 1.577727360831426e-05,
      "loss": 0.0441,
      "step": 4576,
      "task_loss": 0.1002790629863739
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7904687444137731,
      "compression_loss": 0.0,
      "distillation_loss": 0.024910183623433113,
      "epoch": 4.35,
      "learning_rate": 1.5767370763022694e-05,
      "loss": 0.0315,
      "step": 4577,
      "task_loss": 0.09050406515598297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7904761732954233,
      "compression_loss": 0.0,
      "distillation_loss": 0.07750061899423599,
      "epoch": 4.35,
      "learning_rate": 1.5757469594882692e-05,
      "loss": 0.0824,
      "step": 4578,
      "task_loss": 0.12608473002910614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7904835983159084,
      "compression_loss": 0.0,
      "distillation_loss": 0.03907743841409683,
      "epoch": 4.35,
      "learning_rate": 1.574757010569285e-05,
      "loss": 0.0363,
      "step": 4579,
      "task_loss": 0.011346584185957909
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7904910194762322,
      "compression_loss": 0.0,
      "distillation_loss": 0.04754206910729408,
      "epoch": 4.35,
      "learning_rate": 1.5737672297251464e-05,
      "loss": 0.0501,
      "step": 4580,
      "task_loss": 0.07262822985649109
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7904984367773984,
      "compression_loss": 0.0,
      "distillation_loss": 0.07429815083742142,
      "epoch": 4.35,
      "learning_rate": 1.5727776171356506e-05,
      "loss": 0.0701,
      "step": 4581,
      "task_loss": 0.032718729227781296
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7905058502204106,
      "compression_loss": 0.0,
      "distillation_loss": 0.10426107048988342,
      "epoch": 4.35,
      "learning_rate": 1.571788172980566e-05,
      "loss": 0.1008,
      "step": 4582,
      "task_loss": 0.06919807940721512
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7905132598062725,
      "compression_loss": 0.0,
      "distillation_loss": 0.21805521845817566,
      "epoch": 4.35,
      "learning_rate": 1.5707988974396304e-05,
      "loss": 0.2072,
      "step": 4583,
      "task_loss": 0.10992544889450073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7905206655359879,
      "compression_loss": 0.0,
      "distillation_loss": 0.10667760670185089,
      "epoch": 4.35,
      "learning_rate": 1.56980979069255e-05,
      "loss": 0.1054,
      "step": 4584,
      "task_loss": 0.09397716075181961
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7905280674105603,
      "compression_loss": 0.0,
      "distillation_loss": 0.01430472545325756,
      "epoch": 4.35,
      "learning_rate": 1.568820852919002e-05,
      "loss": 0.0183,
      "step": 4585,
      "task_loss": 0.05472009256482124
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7905354654309934,
      "compression_loss": 0.0,
      "distillation_loss": 0.037967242300510406,
      "epoch": 4.36,
      "learning_rate": 1.5678320842986295e-05,
      "loss": 0.0446,
      "step": 4586,
      "task_loss": 0.10416756570339203
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7905428595982911,
      "compression_loss": 0.0,
      "distillation_loss": 0.07452750205993652,
      "epoch": 4.36,
      "learning_rate": 1.5668434850110493e-05,
      "loss": 0.0828,
      "step": 4587,
      "task_loss": 0.1572713553905487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7905502499134568,
      "compression_loss": 0.0,
      "distillation_loss": 0.018898198381066322,
      "epoch": 4.36,
      "learning_rate": 1.565855055235843e-05,
      "loss": 0.0174,
      "step": 4588,
      "task_loss": 0.004319200292229652
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7905576363774943,
      "compression_loss": 0.0,
      "distillation_loss": 0.017305273562669754,
      "epoch": 4.36,
      "learning_rate": 1.5648667951525653e-05,
      "loss": 0.0243,
      "step": 4589,
      "task_loss": 0.08709577471017838
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7905650189914074,
      "compression_loss": 0.0,
      "distillation_loss": 0.02825682982802391,
      "epoch": 4.36,
      "learning_rate": 1.5638787049407382e-05,
      "loss": 0.035,
      "step": 4590,
      "task_loss": 0.09588130563497543
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7905723977561996,
      "compression_loss": 0.0,
      "distillation_loss": 0.03142907842993736,
      "epoch": 4.36,
      "learning_rate": 1.5628907847798517e-05,
      "loss": 0.036,
      "step": 4591,
      "task_loss": 0.07705745846033096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7905797726728746,
      "compression_loss": 0.0,
      "distillation_loss": 0.040951065719127655,
      "epoch": 4.36,
      "learning_rate": 1.5619030348493684e-05,
      "loss": 0.0562,
      "step": 4592,
      "task_loss": 0.19311881065368652
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7905871437424362,
      "compression_loss": 0.0,
      "distillation_loss": 0.023970093578100204,
      "epoch": 4.36,
      "learning_rate": 1.5609154553287163e-05,
      "loss": 0.032,
      "step": 4593,
      "task_loss": 0.10465425252914429
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7905945109658881,
      "compression_loss": 0.0,
      "distillation_loss": 0.03335358202457428,
      "epoch": 4.36,
      "learning_rate": 1.5599280463972953e-05,
      "loss": 0.0312,
      "step": 4594,
      "task_loss": 0.011852225288748741
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7906018743442338,
      "compression_loss": 0.0,
      "distillation_loss": 0.10270802676677704,
      "epoch": 4.36,
      "learning_rate": 1.558940808234471e-05,
      "loss": 0.1207,
      "step": 4595,
      "task_loss": 0.2826935052871704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.790609233878477,
      "compression_loss": 0.0,
      "distillation_loss": 0.11305034160614014,
      "epoch": 4.36,
      "learning_rate": 1.5579537410195817e-05,
      "loss": 0.1194,
      "step": 4596,
      "task_loss": 0.1764708012342453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7906165895696217,
      "compression_loss": 0.0,
      "distillation_loss": 0.02198610082268715,
      "epoch": 4.37,
      "learning_rate": 1.5569668449319323e-05,
      "loss": 0.0204,
      "step": 4597,
      "task_loss": 0.005980312824249268
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7906239414186712,
      "compression_loss": 0.0,
      "distillation_loss": 0.03549554944038391,
      "epoch": 4.37,
      "learning_rate": 1.5559801201507968e-05,
      "loss": 0.0526,
      "step": 4598,
      "task_loss": 0.20651255548000336
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7906312894266293,
      "compression_loss": 0.0,
      "distillation_loss": 0.09966300427913666,
      "epoch": 4.37,
      "learning_rate": 1.554993566855418e-05,
      "loss": 0.1053,
      "step": 4599,
      "task_loss": 0.156356543302536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7906386335944998,
      "compression_loss": 0.0,
      "distillation_loss": 0.0843530148267746,
      "epoch": 4.37,
      "learning_rate": 1.5540071852250106e-05,
      "loss": 0.0909,
      "step": 4600,
      "task_loss": 0.14952674508094788
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7906459739232862,
      "compression_loss": 0.0,
      "distillation_loss": 0.08812104910612106,
      "epoch": 4.37,
      "learning_rate": 1.5530209754387537e-05,
      "loss": 0.0843,
      "step": 4601,
      "task_loss": 0.04979713633656502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7906533104139923,
      "compression_loss": 0.0,
      "distillation_loss": 0.0816584974527359,
      "epoch": 4.37,
      "learning_rate": 1.552034937675797e-05,
      "loss": 0.0981,
      "step": 4602,
      "task_loss": 0.24617856740951538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7906606430676218,
      "compression_loss": 0.0,
      "distillation_loss": 0.022141676396131516,
      "epoch": 4.37,
      "learning_rate": 1.5510490721152592e-05,
      "loss": 0.0206,
      "step": 4603,
      "task_loss": 0.006470389664173126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7906679718851783,
      "compression_loss": 0.0,
      "distillation_loss": 0.039304304867982864,
      "epoch": 4.37,
      "learning_rate": 1.550063378936228e-05,
      "loss": 0.0401,
      "step": 4604,
      "task_loss": 0.047406185418367386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7906752968676656,
      "compression_loss": 0.0,
      "distillation_loss": 0.03900455683469772,
      "epoch": 4.37,
      "learning_rate": 1.549077858317759e-05,
      "loss": 0.0587,
      "step": 4605,
      "task_loss": 0.23572197556495667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7906826180160873,
      "compression_loss": 0.0,
      "distillation_loss": 0.01960546150803566,
      "epoch": 4.37,
      "learning_rate": 1.5480925104388762e-05,
      "loss": 0.0222,
      "step": 4606,
      "task_loss": 0.045594025403261185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.790689935331447,
      "compression_loss": 0.0,
      "distillation_loss": 0.011018482968211174,
      "epoch": 4.38,
      "learning_rate": 1.547107335478574e-05,
      "loss": 0.0162,
      "step": 4607,
      "task_loss": 0.06283082067966461
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7906972488147486,
      "compression_loss": 0.0,
      "distillation_loss": 0.13586799800395966,
      "epoch": 4.38,
      "learning_rate": 1.5461223336158127e-05,
      "loss": 0.1389,
      "step": 4608,
      "task_loss": 0.1658676415681839
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7907045584669955,
      "compression_loss": 0.0,
      "distillation_loss": 0.061904579401016235,
      "epoch": 4.38,
      "learning_rate": 1.5451375050295235e-05,
      "loss": 0.0694,
      "step": 4609,
      "task_loss": 0.1364220827817917
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7907118642891916,
      "compression_loss": 0.0,
      "distillation_loss": 0.06705223023891449,
      "epoch": 4.38,
      "learning_rate": 1.5441528498986053e-05,
      "loss": 0.0632,
      "step": 4610,
      "task_loss": 0.028953341767191887
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7907191662823405,
      "compression_loss": 0.0,
      "distillation_loss": 0.06535607576370239,
      "epoch": 4.38,
      "learning_rate": 1.543168368401926e-05,
      "loss": 0.0687,
      "step": 4611,
      "task_loss": 0.09915080666542053
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7907264644474459,
      "compression_loss": 0.0,
      "distillation_loss": 0.041197728365659714,
      "epoch": 4.38,
      "learning_rate": 1.5421840607183203e-05,
      "loss": 0.0381,
      "step": 4612,
      "task_loss": 0.009746959432959557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7907337587855114,
      "compression_loss": 0.0,
      "distillation_loss": 0.009639294818043709,
      "epoch": 4.38,
      "learning_rate": 1.5411999270265924e-05,
      "loss": 0.0091,
      "step": 4613,
      "task_loss": 0.004122687503695488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.790741049297541,
      "compression_loss": 0.0,
      "distillation_loss": 0.04923541843891144,
      "epoch": 4.38,
      "learning_rate": 1.5402159675055166e-05,
      "loss": 0.0499,
      "step": 4614,
      "task_loss": 0.05560116469860077
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.790748335984538,
      "compression_loss": 0.0,
      "distillation_loss": 0.062305059283971786,
      "epoch": 4.38,
      "learning_rate": 1.5392321823338318e-05,
      "loss": 0.0646,
      "step": 4615,
      "task_loss": 0.08488155901432037
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7907556188475061,
      "compression_loss": 0.0,
      "distillation_loss": 0.08541512489318848,
      "epoch": 4.38,
      "learning_rate": 1.5382485716902486e-05,
      "loss": 0.0906,
      "step": 4616,
      "task_loss": 0.13751107454299927
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7907628978874492,
      "compression_loss": 0.0,
      "distillation_loss": 0.01606060564517975,
      "epoch": 4.38,
      "learning_rate": 1.537265135753443e-05,
      "loss": 0.0216,
      "step": 4617,
      "task_loss": 0.07104544341564178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.790770173105371,
      "compression_loss": 0.0,
      "distillation_loss": 0.013350674882531166,
      "epoch": 4.39,
      "learning_rate": 1.536281874702063e-05,
      "loss": 0.0173,
      "step": 4618,
      "task_loss": 0.05334080010652542
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7907774445022749,
      "compression_loss": 0.0,
      "distillation_loss": 0.03925604373216629,
      "epoch": 4.39,
      "learning_rate": 1.535298788714722e-05,
      "loss": 0.0443,
      "step": 4619,
      "task_loss": 0.08980512619018555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7907847120791649,
      "compression_loss": 0.0,
      "distillation_loss": 0.07377579063177109,
      "epoch": 4.39,
      "learning_rate": 1.5343158779700016e-05,
      "loss": 0.0715,
      "step": 4620,
      "task_loss": 0.050635963678359985
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7907919758370445,
      "compression_loss": 0.0,
      "distillation_loss": 0.034185461699962616,
      "epoch": 4.39,
      "learning_rate": 1.533333142646453e-05,
      "loss": 0.0327,
      "step": 4621,
      "task_loss": 0.019142286852002144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7907992357769174,
      "compression_loss": 0.0,
      "distillation_loss": 0.08377894014120102,
      "epoch": 4.39,
      "learning_rate": 1.5323505829225947e-05,
      "loss": 0.0824,
      "step": 4622,
      "task_loss": 0.07016059011220932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7908064918997872,
      "compression_loss": 0.0,
      "distillation_loss": 0.13712866604328156,
      "epoch": 4.39,
      "learning_rate": 1.5313681989769136e-05,
      "loss": 0.1362,
      "step": 4623,
      "task_loss": 0.12798915803432465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7908137442066578,
      "compression_loss": 0.0,
      "distillation_loss": 0.08357212692499161,
      "epoch": 4.39,
      "learning_rate": 1.530385990987863e-05,
      "loss": 0.0788,
      "step": 4624,
      "task_loss": 0.035898420959711075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7908209926985328,
      "compression_loss": 0.0,
      "distillation_loss": 0.025412458926439285,
      "epoch": 4.39,
      "learning_rate": 1.529403959133867e-05,
      "loss": 0.0407,
      "step": 4625,
      "task_loss": 0.17801359295845032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7908282373764158,
      "compression_loss": 0.0,
      "distillation_loss": 0.020880520343780518,
      "epoch": 4.39,
      "learning_rate": 1.5284221035933166e-05,
      "loss": 0.0194,
      "step": 4626,
      "task_loss": 0.006218817085027695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7908354782413105,
      "compression_loss": 0.0,
      "distillation_loss": 0.09893855452537537,
      "epoch": 4.39,
      "learning_rate": 1.5274404245445704e-05,
      "loss": 0.0946,
      "step": 4627,
      "task_loss": 0.05531448870897293
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7908427152942207,
      "compression_loss": 0.0,
      "distillation_loss": 0.13214893639087677,
      "epoch": 4.4,
      "learning_rate": 1.5264589221659553e-05,
      "loss": 0.1336,
      "step": 4628,
      "task_loss": 0.14715173840522766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7908499485361499,
      "compression_loss": 0.0,
      "distillation_loss": 0.11516699194908142,
      "epoch": 4.4,
      "learning_rate": 1.5254775966357653e-05,
      "loss": 0.1095,
      "step": 4629,
      "task_loss": 0.058957893401384354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7908571779681018,
      "compression_loss": 0.0,
      "distillation_loss": 0.08969461172819138,
      "epoch": 4.4,
      "learning_rate": 1.5244964481322637e-05,
      "loss": 0.0989,
      "step": 4630,
      "task_loss": 0.18196004629135132
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7908644035910803,
      "compression_loss": 0.0,
      "distillation_loss": 0.1797245740890503,
      "epoch": 4.4,
      "learning_rate": 1.5235154768336795e-05,
      "loss": 0.1714,
      "step": 4631,
      "task_loss": 0.09689280390739441
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7908716254060889,
      "compression_loss": 0.0,
      "distillation_loss": 0.08713481575250626,
      "epoch": 4.4,
      "learning_rate": 1.5225346829182121e-05,
      "loss": 0.0895,
      "step": 4632,
      "task_loss": 0.1106729805469513
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7908788434141313,
      "compression_loss": 0.0,
      "distillation_loss": 0.05905129015445709,
      "epoch": 4.4,
      "learning_rate": 1.5215540665640277e-05,
      "loss": 0.0611,
      "step": 4633,
      "task_loss": 0.07935148477554321
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7908860576162112,
      "compression_loss": 0.0,
      "distillation_loss": 0.08967553079128265,
      "epoch": 4.4,
      "learning_rate": 1.5205736279492574e-05,
      "loss": 0.0875,
      "step": 4634,
      "task_loss": 0.0675487220287323
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7908932680133324,
      "compression_loss": 0.0,
      "distillation_loss": 0.07612395286560059,
      "epoch": 4.4,
      "learning_rate": 1.5195933672520064e-05,
      "loss": 0.0825,
      "step": 4635,
      "task_loss": 0.13981005549430847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7909004746064983,
      "compression_loss": 0.0,
      "distillation_loss": 0.18954381346702576,
      "epoch": 4.4,
      "learning_rate": 1.5186132846503412e-05,
      "loss": 0.1939,
      "step": 4636,
      "task_loss": 0.23316198587417603
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7909076773967129,
      "compression_loss": 0.0,
      "distillation_loss": 0.02852441743016243,
      "epoch": 4.4,
      "learning_rate": 1.5176333803222998e-05,
      "loss": 0.0353,
      "step": 4637,
      "task_loss": 0.09584514796733856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7909148763849796,
      "compression_loss": 0.0,
      "distillation_loss": 0.04975371062755585,
      "epoch": 4.4,
      "learning_rate": 1.5166536544458856e-05,
      "loss": 0.0526,
      "step": 4638,
      "task_loss": 0.0777263194322586
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7909220715723022,
      "compression_loss": 0.0,
      "distillation_loss": 0.17351922392845154,
      "epoch": 4.41,
      "learning_rate": 1.515674107199071e-05,
      "loss": 0.1641,
      "step": 4639,
      "task_loss": 0.07936342060565948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7909292629596845,
      "compression_loss": 0.0,
      "distillation_loss": 0.028591414913535118,
      "epoch": 4.41,
      "learning_rate": 1.5146947387597956e-05,
      "loss": 0.0428,
      "step": 4640,
      "task_loss": 0.17051829397678375
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79093645054813,
      "compression_loss": 0.0,
      "distillation_loss": 0.03785558417439461,
      "epoch": 4.41,
      "learning_rate": 1.513715549305966e-05,
      "loss": 0.0441,
      "step": 4641,
      "task_loss": 0.10044631361961365
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7909436343386426,
      "compression_loss": 0.0,
      "distillation_loss": 0.09519485384225845,
      "epoch": 4.41,
      "learning_rate": 1.512736539015457e-05,
      "loss": 0.0916,
      "step": 4642,
      "task_loss": 0.05921884626150131
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7909508143322257,
      "compression_loss": 0.0,
      "distillation_loss": 0.034322589635849,
      "epoch": 4.41,
      "learning_rate": 1.5117577080661094e-05,
      "loss": 0.0324,
      "step": 4643,
      "task_loss": 0.014751961454749107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7909579905298831,
      "compression_loss": 0.0,
      "distillation_loss": 0.01889497973024845,
      "epoch": 4.41,
      "learning_rate": 1.5107790566357347e-05,
      "loss": 0.0176,
      "step": 4644,
      "task_loss": 0.005802595987915993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7909651629326186,
      "compression_loss": 0.0,
      "distillation_loss": 0.09566991776227951,
      "epoch": 4.41,
      "learning_rate": 1.509800584902108e-05,
      "loss": 0.0992,
      "step": 4645,
      "task_loss": 0.13102050125598907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7909723315414358,
      "compression_loss": 0.0,
      "distillation_loss": 0.06106077879667282,
      "epoch": 4.41,
      "learning_rate": 1.508822293042974e-05,
      "loss": 0.0612,
      "step": 4646,
      "task_loss": 0.06258445233106613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7909794963573382,
      "compression_loss": 0.0,
      "distillation_loss": 0.046314168721437454,
      "epoch": 4.41,
      "learning_rate": 1.5078441812360445e-05,
      "loss": 0.0432,
      "step": 4647,
      "task_loss": 0.01564324088394642
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7909866573813299,
      "compression_loss": 0.0,
      "distillation_loss": 0.05536476522684097,
      "epoch": 4.41,
      "learning_rate": 1.5068662496589975e-05,
      "loss": 0.0511,
      "step": 4648,
      "task_loss": 0.012430863454937935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7909938146144142,
      "compression_loss": 0.0,
      "distillation_loss": 0.08107525110244751,
      "epoch": 4.42,
      "learning_rate": 1.5058884984894788e-05,
      "loss": 0.0793,
      "step": 4649,
      "task_loss": 0.06296955794095993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7910009680575949,
      "compression_loss": 0.0,
      "distillation_loss": 0.04533292353153229,
      "epoch": 4.42,
      "learning_rate": 1.5049109279051026e-05,
      "loss": 0.0599,
      "step": 4650,
      "task_loss": 0.1911742091178894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7910081177118757,
      "compression_loss": 0.0,
      "distillation_loss": 0.05968482419848442,
      "epoch": 4.42,
      "learning_rate": 1.503933538083448e-05,
      "loss": 0.0574,
      "step": 4651,
      "task_loss": 0.03725713863968849
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7910152635782604,
      "compression_loss": 0.0,
      "distillation_loss": 0.04093274474143982,
      "epoch": 4.42,
      "learning_rate": 1.5029563292020637e-05,
      "loss": 0.0498,
      "step": 4652,
      "task_loss": 0.12984903156757355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7910224056577524,
      "compression_loss": 0.0,
      "distillation_loss": 0.0636286810040474,
      "epoch": 4.42,
      "learning_rate": 1.5019793014384643e-05,
      "loss": 0.0631,
      "step": 4653,
      "task_loss": 0.05869031697511673
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7910295439513556,
      "compression_loss": 0.0,
      "distillation_loss": 0.11052249372005463,
      "epoch": 4.42,
      "learning_rate": 1.5010024549701312e-05,
      "loss": 0.1109,
      "step": 4654,
      "task_loss": 0.11410997062921524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7910366784600736,
      "compression_loss": 0.0,
      "distillation_loss": 0.06556328386068344,
      "epoch": 4.42,
      "learning_rate": 1.5000257899745134e-05,
      "loss": 0.0636,
      "step": 4655,
      "task_loss": 0.04608140140771866
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7910438091849102,
      "compression_loss": 0.0,
      "distillation_loss": 0.027359087020158768,
      "epoch": 4.42,
      "learning_rate": 1.4990493066290265e-05,
      "loss": 0.0253,
      "step": 4656,
      "task_loss": 0.0065147485584020615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7910509361268689,
      "compression_loss": 0.0,
      "distillation_loss": 0.03795522451400757,
      "epoch": 4.42,
      "learning_rate": 1.4980730051110541e-05,
      "loss": 0.0557,
      "step": 4657,
      "task_loss": 0.21516427397727966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7910580592869536,
      "compression_loss": 0.0,
      "distillation_loss": 0.03814719617366791,
      "epoch": 4.42,
      "learning_rate": 1.4970968855979455e-05,
      "loss": 0.046,
      "step": 4658,
      "task_loss": 0.11639110743999481
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7910651786661678,
      "compression_loss": 0.0,
      "distillation_loss": 0.06086570769548416,
      "epoch": 4.42,
      "learning_rate": 1.496120948267018e-05,
      "loss": 0.0621,
      "step": 4659,
      "task_loss": 0.072751984000206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7910722942655152,
      "compression_loss": 0.0,
      "distillation_loss": 0.02917177602648735,
      "epoch": 4.43,
      "learning_rate": 1.4951451932955534e-05,
      "loss": 0.0277,
      "step": 4660,
      "task_loss": 0.014873206615447998
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7910794060859996,
      "compression_loss": 0.0,
      "distillation_loss": 0.05168045312166214,
      "epoch": 4.43,
      "learning_rate": 1.4941696208608056e-05,
      "loss": 0.0492,
      "step": 4661,
      "task_loss": 0.027083786204457283
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7910865141286245,
      "compression_loss": 0.0,
      "distillation_loss": 0.09697671234607697,
      "epoch": 4.43,
      "learning_rate": 1.4931942311399896e-05,
      "loss": 0.1167,
      "step": 4662,
      "task_loss": 0.29466700553894043
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7910936183943937,
      "compression_loss": 0.0,
      "distillation_loss": 0.09114208817481995,
      "epoch": 4.43,
      "learning_rate": 1.4922190243102905e-05,
      "loss": 0.0878,
      "step": 4663,
      "task_loss": 0.057473860681056976
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7911007188843109,
      "compression_loss": 0.0,
      "distillation_loss": 0.05134554207324982,
      "epoch": 4.43,
      "learning_rate": 1.4912440005488593e-05,
      "loss": 0.0732,
      "step": 4664,
      "task_loss": 0.2698168158531189
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7911078155993797,
      "compression_loss": 0.0,
      "distillation_loss": 0.024719703942537308,
      "epoch": 4.43,
      "learning_rate": 1.4902691600328134e-05,
      "loss": 0.023,
      "step": 4665,
      "task_loss": 0.0074704308062791824
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7911149085406038,
      "compression_loss": 0.0,
      "distillation_loss": 0.11209557205438614,
      "epoch": 4.43,
      "learning_rate": 1.489294502939238e-05,
      "loss": 0.1162,
      "step": 4666,
      "task_loss": 0.15347597002983093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7911219977089871,
      "compression_loss": 0.0,
      "distillation_loss": 0.08372636884450912,
      "epoch": 4.43,
      "learning_rate": 1.4883200294451832e-05,
      "loss": 0.0796,
      "step": 4667,
      "task_loss": 0.04254281520843506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7911290831055329,
      "compression_loss": 0.0,
      "distillation_loss": 0.045971132814884186,
      "epoch": 4.43,
      "learning_rate": 1.4873457397276675e-05,
      "loss": 0.0433,
      "step": 4668,
      "task_loss": 0.0188460536301136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7911361647312452,
      "compression_loss": 0.0,
      "distillation_loss": 0.012892801314592361,
      "epoch": 4.43,
      "learning_rate": 1.4863716339636746e-05,
      "loss": 0.0171,
      "step": 4669,
      "task_loss": 0.05512375757098198
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7911432425871274,
      "compression_loss": 0.0,
      "distillation_loss": 0.05050332099199295,
      "epoch": 4.43,
      "learning_rate": 1.4853977123301565e-05,
      "loss": 0.062,
      "step": 4670,
      "task_loss": 0.16584210097789764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7911503166741836,
      "compression_loss": 0.0,
      "distillation_loss": 0.030305128544569016,
      "epoch": 4.44,
      "learning_rate": 1.4844239750040308e-05,
      "loss": 0.0305,
      "step": 4671,
      "task_loss": 0.0323617160320282
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.791157386993417,
      "compression_loss": 0.0,
      "distillation_loss": 0.06064460799098015,
      "epoch": 4.44,
      "learning_rate": 1.483450422162181e-05,
      "loss": 0.0614,
      "step": 4672,
      "task_loss": 0.06825106590986252
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7911644535458315,
      "compression_loss": 0.0,
      "distillation_loss": 0.040299929678440094,
      "epoch": 4.44,
      "learning_rate": 1.4824770539814575e-05,
      "loss": 0.0375,
      "step": 4673,
      "task_loss": 0.012086659669876099
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7911715163324309,
      "compression_loss": 0.0,
      "distillation_loss": 0.05591391772031784,
      "epoch": 4.44,
      "learning_rate": 1.4815038706386777e-05,
      "loss": 0.06,
      "step": 4674,
      "task_loss": 0.09698733687400818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7911785753542188,
      "compression_loss": 0.0,
      "distillation_loss": 0.038244374096393585,
      "epoch": 4.44,
      "learning_rate": 1.4805308723106248e-05,
      "loss": 0.0403,
      "step": 4675,
      "task_loss": 0.0591559074819088
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7911856306121987,
      "compression_loss": 0.0,
      "distillation_loss": 0.0506531223654747,
      "epoch": 4.44,
      "learning_rate": 1.4795580591740493e-05,
      "loss": 0.0497,
      "step": 4676,
      "task_loss": 0.04083885997533798
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7911926821073746,
      "compression_loss": 0.0,
      "distillation_loss": 0.08757050335407257,
      "epoch": 4.44,
      "learning_rate": 1.4785854314056652e-05,
      "loss": 0.0904,
      "step": 4677,
      "task_loss": 0.1154625192284584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7911997298407498,
      "compression_loss": 0.0,
      "distillation_loss": 0.10649511963129044,
      "epoch": 4.44,
      "learning_rate": 1.4776129891821583e-05,
      "loss": 0.1165,
      "step": 4678,
      "task_loss": 0.2064911276102066
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7912067738133284,
      "compression_loss": 0.0,
      "distillation_loss": 0.06318517029285431,
      "epoch": 4.44,
      "learning_rate": 1.4766407326801751e-05,
      "loss": 0.0593,
      "step": 4679,
      "task_loss": 0.02464883401989937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7912138140261138,
      "compression_loss": 0.0,
      "distillation_loss": 0.020300816744565964,
      "epoch": 4.44,
      "learning_rate": 1.4756686620763322e-05,
      "loss": 0.0197,
      "step": 4680,
      "task_loss": 0.013965649530291557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7912208504801098,
      "compression_loss": 0.0,
      "distillation_loss": 0.01607617177069187,
      "epoch": 4.45,
      "learning_rate": 1.4746967775472093e-05,
      "loss": 0.0226,
      "step": 4681,
      "task_loss": 0.08178968727588654
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79122788317632,
      "compression_loss": 0.0,
      "distillation_loss": 0.09621655941009521,
      "epoch": 4.45,
      "learning_rate": 1.4737250792693546e-05,
      "loss": 0.0925,
      "step": 4682,
      "task_loss": 0.05869336053729057
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7912349121157481,
      "compression_loss": 0.0,
      "distillation_loss": 0.02147580124437809,
      "epoch": 4.45,
      "learning_rate": 1.4727535674192825e-05,
      "loss": 0.0198,
      "step": 4683,
      "task_loss": 0.004298372194170952
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7912419372993978,
      "compression_loss": 0.0,
      "distillation_loss": 0.041834309697151184,
      "epoch": 4.45,
      "learning_rate": 1.4717822421734718e-05,
      "loss": 0.0438,
      "step": 4684,
      "task_loss": 0.06110651418566704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7912489587282729,
      "compression_loss": 0.0,
      "distillation_loss": 0.045896559953689575,
      "epoch": 4.45,
      "learning_rate": 1.4708111037083683e-05,
      "loss": 0.0442,
      "step": 4685,
      "task_loss": 0.029097534716129303
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7912559764033769,
      "compression_loss": 0.0,
      "distillation_loss": 0.030547471717000008,
      "epoch": 4.45,
      "learning_rate": 1.4698401522003843e-05,
      "loss": 0.0281,
      "step": 4686,
      "task_loss": 0.006017416715621948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7912629903257136,
      "compression_loss": 0.0,
      "distillation_loss": 0.022139865905046463,
      "epoch": 4.45,
      "learning_rate": 1.4688693878258991e-05,
      "loss": 0.0209,
      "step": 4687,
      "task_loss": 0.009355850517749786
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7912700004962866,
      "compression_loss": 0.0,
      "distillation_loss": 0.029008762910962105,
      "epoch": 4.45,
      "learning_rate": 1.4678988107612546e-05,
      "loss": 0.0366,
      "step": 4688,
      "task_loss": 0.10471072793006897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7912770069160996,
      "compression_loss": 0.0,
      "distillation_loss": 0.019571123644709587,
      "epoch": 4.45,
      "learning_rate": 1.4669284211827622e-05,
      "loss": 0.0302,
      "step": 4689,
      "task_loss": 0.12538588047027588
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7912840095861563,
      "compression_loss": 0.0,
      "distillation_loss": 0.08734140545129776,
      "epoch": 4.45,
      "learning_rate": 1.4659582192666977e-05,
      "loss": 0.0883,
      "step": 4690,
      "task_loss": 0.09734951704740524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7912910085074604,
      "compression_loss": 0.0,
      "distillation_loss": 0.021236347034573555,
      "epoch": 4.45,
      "learning_rate": 1.4649882051893022e-05,
      "loss": 0.0276,
      "step": 4691,
      "task_loss": 0.08498524129390717
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7912980036810155,
      "compression_loss": 0.0,
      "distillation_loss": 0.07887224853038788,
      "epoch": 4.46,
      "learning_rate": 1.464018379126784e-05,
      "loss": 0.0804,
      "step": 4692,
      "task_loss": 0.09387266635894775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7913049951078255,
      "compression_loss": 0.0,
      "distillation_loss": 0.06495728343725204,
      "epoch": 4.46,
      "learning_rate": 1.4630487412553168e-05,
      "loss": 0.0835,
      "step": 4693,
      "task_loss": 0.2504326403141022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7913119827888938,
      "compression_loss": 0.0,
      "distillation_loss": 0.0405736081302166,
      "epoch": 4.46,
      "learning_rate": 1.4620792917510395e-05,
      "loss": 0.06,
      "step": 4694,
      "task_loss": 0.2343919277191162
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7913189667252243,
      "compression_loss": 0.0,
      "distillation_loss": 0.05520131438970566,
      "epoch": 4.46,
      "learning_rate": 1.4611100307900572e-05,
      "loss": 0.0586,
      "step": 4695,
      "task_loss": 0.08893582224845886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7913259469178205,
      "compression_loss": 0.0,
      "distillation_loss": 0.05112463980913162,
      "epoch": 4.46,
      "learning_rate": 1.4601409585484413e-05,
      "loss": 0.0487,
      "step": 4696,
      "task_loss": 0.02662830613553524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7913329233676862,
      "compression_loss": 0.0,
      "distillation_loss": 0.0339503176510334,
      "epoch": 4.46,
      "learning_rate": 1.4591720752022286e-05,
      "loss": 0.0313,
      "step": 4697,
      "task_loss": 0.007592087611556053
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7913398960758251,
      "compression_loss": 0.0,
      "distillation_loss": 0.01650651916861534,
      "epoch": 4.46,
      "learning_rate": 1.4582033809274215e-05,
      "loss": 0.0153,
      "step": 4698,
      "task_loss": 0.004532504826784134
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7913468650432408,
      "compression_loss": 0.0,
      "distillation_loss": 0.08176242560148239,
      "epoch": 4.46,
      "learning_rate": 1.4572348758999877e-05,
      "loss": 0.0844,
      "step": 4699,
      "task_loss": 0.10765102505683899
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7913538302709371,
      "compression_loss": 0.0,
      "distillation_loss": 0.036356210708618164,
      "epoch": 4.46,
      "learning_rate": 1.4562665602958592e-05,
      "loss": 0.0466,
      "step": 4700,
      "task_loss": 0.13854879140853882
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7913607917599175,
      "compression_loss": 0.0,
      "distillation_loss": 0.14445237815380096,
      "epoch": 4.46,
      "learning_rate": 1.4552984342909382e-05,
      "loss": 0.1413,
      "step": 4701,
      "task_loss": 0.11329984664916992
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7913677495111858,
      "compression_loss": 0.0,
      "distillation_loss": 0.015349620953202248,
      "epoch": 4.47,
      "learning_rate": 1.4543304980610878e-05,
      "loss": 0.0141,
      "step": 4702,
      "task_loss": 0.003350917249917984
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7913747035257457,
      "compression_loss": 0.0,
      "distillation_loss": 0.11373439431190491,
      "epoch": 4.47,
      "learning_rate": 1.4533627517821374e-05,
      "loss": 0.127,
      "step": 4703,
      "task_loss": 0.24672245979309082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7913816538046009,
      "compression_loss": 0.0,
      "distillation_loss": 0.09408354014158249,
      "epoch": 4.47,
      "learning_rate": 1.452395195629884e-05,
      "loss": 0.0891,
      "step": 4704,
      "task_loss": 0.04441932588815689
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7913886003487549,
      "compression_loss": 0.0,
      "distillation_loss": 0.09586971998214722,
      "epoch": 4.47,
      "learning_rate": 1.4514278297800893e-05,
      "loss": 0.0924,
      "step": 4705,
      "task_loss": 0.06134894862771034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7913955431592116,
      "compression_loss": 0.0,
      "distillation_loss": 0.09412407875061035,
      "epoch": 4.47,
      "learning_rate": 1.4504606544084798e-05,
      "loss": 0.0965,
      "step": 4706,
      "task_loss": 0.1174854040145874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7914024822369746,
      "compression_loss": 0.0,
      "distillation_loss": 0.07334903627634048,
      "epoch": 4.47,
      "learning_rate": 1.4494936696907458e-05,
      "loss": 0.0821,
      "step": 4707,
      "task_loss": 0.1608203798532486
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7914094175830476,
      "compression_loss": 0.0,
      "distillation_loss": 0.023582283407449722,
      "epoch": 4.47,
      "learning_rate": 1.4485268758025466e-05,
      "loss": 0.0217,
      "step": 4708,
      "task_loss": 0.0046432409435510635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7914163491984342,
      "compression_loss": 0.0,
      "distillation_loss": 0.06829246878623962,
      "epoch": 4.47,
      "learning_rate": 1.4475602729195048e-05,
      "loss": 0.0707,
      "step": 4709,
      "task_loss": 0.09261967241764069
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7914232770841382,
      "compression_loss": 0.0,
      "distillation_loss": 0.05515692010521889,
      "epoch": 4.47,
      "learning_rate": 1.446593861217207e-05,
      "loss": 0.059,
      "step": 4710,
      "task_loss": 0.09347251802682877
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7914302012411631,
      "compression_loss": 0.0,
      "distillation_loss": 0.0656106099486351,
      "epoch": 4.47,
      "learning_rate": 1.4456276408712083e-05,
      "loss": 0.0618,
      "step": 4711,
      "task_loss": 0.027708284556865692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7914371216705128,
      "compression_loss": 0.0,
      "distillation_loss": 0.02283991128206253,
      "epoch": 4.47,
      "learning_rate": 1.4446616120570258e-05,
      "loss": 0.0257,
      "step": 4712,
      "task_loss": 0.051784027367830276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7914440383731909,
      "compression_loss": 0.0,
      "distillation_loss": 0.11995580047369003,
      "epoch": 4.48,
      "learning_rate": 1.443695774950145e-05,
      "loss": 0.1159,
      "step": 4713,
      "task_loss": 0.07978808879852295
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.791450951350201,
      "compression_loss": 0.0,
      "distillation_loss": 0.16006718575954437,
      "epoch": 4.48,
      "learning_rate": 1.4427301297260129e-05,
      "loss": 0.1553,
      "step": 4714,
      "task_loss": 0.1124170571565628
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.791457860602547,
      "compression_loss": 0.0,
      "distillation_loss": 0.09785357117652893,
      "epoch": 4.48,
      "learning_rate": 1.4417646765600457e-05,
      "loss": 0.0959,
      "step": 4715,
      "task_loss": 0.07808717340230942
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7914647661312323,
      "compression_loss": 0.0,
      "distillation_loss": 0.06697249412536621,
      "epoch": 4.48,
      "learning_rate": 1.4407994156276212e-05,
      "loss": 0.0734,
      "step": 4716,
      "task_loss": 0.131184920668602
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7914716679372608,
      "compression_loss": 0.0,
      "distillation_loss": 0.04526514932513237,
      "epoch": 4.48,
      "learning_rate": 1.4398343471040831e-05,
      "loss": 0.0416,
      "step": 4717,
      "task_loss": 0.009023293852806091
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.791478566021636,
      "compression_loss": 0.0,
      "distillation_loss": 0.03617861866950989,
      "epoch": 4.48,
      "learning_rate": 1.438869471164743e-05,
      "loss": 0.0367,
      "step": 4718,
      "task_loss": 0.04139818996191025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7914854603853618,
      "compression_loss": 0.0,
      "distillation_loss": 0.013502972200512886,
      "epoch": 4.48,
      "learning_rate": 1.4379047879848736e-05,
      "loss": 0.0128,
      "step": 4719,
      "task_loss": 0.006139175966382027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7914923510294417,
      "compression_loss": 0.0,
      "distillation_loss": 0.03151702880859375,
      "epoch": 4.48,
      "learning_rate": 1.4369402977397148e-05,
      "loss": 0.0295,
      "step": 4720,
      "task_loss": 0.011361634358763695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7914992379548794,
      "compression_loss": 0.0,
      "distillation_loss": 0.08465395122766495,
      "epoch": 4.48,
      "learning_rate": 1.4359760006044686e-05,
      "loss": 0.0813,
      "step": 4721,
      "task_loss": 0.051406532526016235
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7915061211626787,
      "compression_loss": 0.0,
      "distillation_loss": 0.02832719124853611,
      "epoch": 4.48,
      "learning_rate": 1.435011896754308e-05,
      "loss": 0.0412,
      "step": 4722,
      "task_loss": 0.15668340027332306
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7915130006538432,
      "compression_loss": 0.0,
      "distillation_loss": 0.05229802057147026,
      "epoch": 4.49,
      "learning_rate": 1.4340479863643658e-05,
      "loss": 0.0571,
      "step": 4723,
      "task_loss": 0.10068703442811966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7915198764293766,
      "compression_loss": 0.0,
      "distillation_loss": 0.019345447421073914,
      "epoch": 4.49,
      "learning_rate": 1.4330842696097393e-05,
      "loss": 0.018,
      "step": 4724,
      "task_loss": 0.006194958463311195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7915267484902825,
      "compression_loss": 0.0,
      "distillation_loss": 0.08192586153745651,
      "epoch": 4.49,
      "learning_rate": 1.4321207466654945e-05,
      "loss": 0.0811,
      "step": 4725,
      "task_loss": 0.07349298149347305
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7915336168375647,
      "compression_loss": 0.0,
      "distillation_loss": 0.12662634253501892,
      "epoch": 4.49,
      "learning_rate": 1.4311574177066594e-05,
      "loss": 0.1275,
      "step": 4726,
      "task_loss": 0.1349095106124878
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7915404814722269,
      "compression_loss": 0.0,
      "distillation_loss": 0.01997673325240612,
      "epoch": 4.49,
      "learning_rate": 1.4301942829082265e-05,
      "loss": 0.0185,
      "step": 4727,
      "task_loss": 0.005590014159679413
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7915473423952726,
      "compression_loss": 0.0,
      "distillation_loss": 0.025746600702404976,
      "epoch": 4.49,
      "learning_rate": 1.4292313424451536e-05,
      "loss": 0.024,
      "step": 4728,
      "task_loss": 0.007853610441088676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7915541996077057,
      "compression_loss": 0.0,
      "distillation_loss": 0.03533206880092621,
      "epoch": 4.49,
      "learning_rate": 1.4282685964923642e-05,
      "loss": 0.0372,
      "step": 4729,
      "task_loss": 0.05367380380630493
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7915610531105297,
      "compression_loss": 0.0,
      "distillation_loss": 0.06636801362037659,
      "epoch": 4.49,
      "learning_rate": 1.427306045224747e-05,
      "loss": 0.0671,
      "step": 4730,
      "task_loss": 0.07363536953926086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7915679029047484,
      "compression_loss": 0.0,
      "distillation_loss": 0.15005621314048767,
      "epoch": 4.49,
      "learning_rate": 1.4263436888171516e-05,
      "loss": 0.1441,
      "step": 4731,
      "task_loss": 0.0903296247124672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7915747489913655,
      "compression_loss": 0.0,
      "distillation_loss": 0.14973436295986176,
      "epoch": 4.49,
      "learning_rate": 1.4253815274443965e-05,
      "loss": 0.1512,
      "step": 4732,
      "task_loss": 0.16403284668922424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7915815913713846,
      "compression_loss": 0.0,
      "distillation_loss": 0.04621317982673645,
      "epoch": 4.49,
      "learning_rate": 1.424419561281263e-05,
      "loss": 0.0627,
      "step": 4733,
      "task_loss": 0.21088391542434692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7915884300458095,
      "compression_loss": 0.0,
      "distillation_loss": 0.021405549719929695,
      "epoch": 4.5,
      "learning_rate": 1.423457790502496e-05,
      "loss": 0.0199,
      "step": 4734,
      "task_loss": 0.0061973873525857925
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7915952650156436,
      "compression_loss": 0.0,
      "distillation_loss": 0.07480795681476593,
      "epoch": 4.5,
      "learning_rate": 1.4224962152828054e-05,
      "loss": 0.0678,
      "step": 4735,
      "task_loss": 0.004553038626909256
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7916020962818909,
      "compression_loss": 0.0,
      "distillation_loss": 0.023463796824216843,
      "epoch": 4.5,
      "learning_rate": 1.4215348357968669e-05,
      "loss": 0.0456,
      "step": 4736,
      "task_loss": 0.24516035616397858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7916089238455549,
      "compression_loss": 0.0,
      "distillation_loss": 0.019841229543089867,
      "epoch": 4.5,
      "learning_rate": 1.4205736522193197e-05,
      "loss": 0.0224,
      "step": 4737,
      "task_loss": 0.0451948307454586
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7916157477076394,
      "compression_loss": 0.0,
      "distillation_loss": 0.11894679069519043,
      "epoch": 4.5,
      "learning_rate": 1.4196126647247654e-05,
      "loss": 0.1265,
      "step": 4738,
      "task_loss": 0.19399774074554443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7916225678691481,
      "compression_loss": 0.0,
      "distillation_loss": 0.023783180862665176,
      "epoch": 4.5,
      "learning_rate": 1.4186518734877757e-05,
      "loss": 0.0354,
      "step": 4739,
      "task_loss": 0.13973920047283173
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7916293843310844,
      "compression_loss": 0.0,
      "distillation_loss": 0.04689887911081314,
      "epoch": 4.5,
      "learning_rate": 1.4176912786828808e-05,
      "loss": 0.0544,
      "step": 4740,
      "task_loss": 0.12143304944038391
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7916361970944523,
      "compression_loss": 0.0,
      "distillation_loss": 0.02539118006825447,
      "epoch": 4.5,
      "learning_rate": 1.4167308804845774e-05,
      "loss": 0.0285,
      "step": 4741,
      "task_loss": 0.05643775314092636
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7916430061602554,
      "compression_loss": 0.0,
      "distillation_loss": 0.04983173683285713,
      "epoch": 4.5,
      "learning_rate": 1.4157706790673262e-05,
      "loss": 0.0469,
      "step": 4742,
      "task_loss": 0.020453961566090584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7916498115294973,
      "compression_loss": 0.0,
      "distillation_loss": 0.023538459092378616,
      "epoch": 4.5,
      "learning_rate": 1.4148106746055535e-05,
      "loss": 0.0218,
      "step": 4743,
      "task_loss": 0.0058768633753061295
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7916566132031818,
      "compression_loss": 0.0,
      "distillation_loss": 0.07629226893186569,
      "epoch": 4.51,
      "learning_rate": 1.4138508672736483e-05,
      "loss": 0.0728,
      "step": 4744,
      "task_loss": 0.04148370400071144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7916634111823124,
      "compression_loss": 0.0,
      "distillation_loss": 0.049477558583021164,
      "epoch": 4.51,
      "learning_rate": 1.4128912572459629e-05,
      "loss": 0.0464,
      "step": 4745,
      "task_loss": 0.018751783296465874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.791670205467893,
      "compression_loss": 0.0,
      "distillation_loss": 0.1168399229645729,
      "epoch": 4.51,
      "learning_rate": 1.4119318446968171e-05,
      "loss": 0.1262,
      "step": 4746,
      "task_loss": 0.21064327657222748
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7916769960609271,
      "compression_loss": 0.0,
      "distillation_loss": 0.03802032768726349,
      "epoch": 4.51,
      "learning_rate": 1.4109726298004911e-05,
      "loss": 0.0437,
      "step": 4747,
      "task_loss": 0.09514382481575012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7916837829624186,
      "compression_loss": 0.0,
      "distillation_loss": 0.05071219056844711,
      "epoch": 4.51,
      "learning_rate": 1.4100136127312324e-05,
      "loss": 0.0538,
      "step": 4748,
      "task_loss": 0.0818718820810318
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7916905661733709,
      "compression_loss": 0.0,
      "distillation_loss": 0.05683235824108124,
      "epoch": 4.51,
      "learning_rate": 1.4090547936632494e-05,
      "loss": 0.0543,
      "step": 4749,
      "task_loss": 0.03133031725883484
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7916973456947879,
      "compression_loss": 0.0,
      "distillation_loss": 0.02370285615324974,
      "epoch": 4.51,
      "learning_rate": 1.4080961727707184e-05,
      "loss": 0.0217,
      "step": 4750,
      "task_loss": 0.0038769226521253586
    },
    {
      "epoch": 4.51,
      "eval_accuracy": 0.8922018348623854,
      "eval_loss": 0.4650850296020508,
      "eval_runtime": 18.1377,
      "eval_samples_per_second": 48.077,
      "eval_steps_per_second": 6.01,
      "step": 4750
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7917041215276731,
      "compression_loss": 0.0,
      "distillation_loss": 0.02110632136464119,
      "epoch": 4.51,
      "learning_rate": 1.4071377502277764e-05,
      "loss": 0.0196,
      "step": 4751,
      "task_loss": 0.005888473242521286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7917108936730304,
      "compression_loss": 0.0,
      "distillation_loss": 0.04167748987674713,
      "epoch": 4.51,
      "learning_rate": 1.4061795262085243e-05,
      "loss": 0.0383,
      "step": 4752,
      "task_loss": 0.007969718426465988
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7917176621318633,
      "compression_loss": 0.0,
      "distillation_loss": 0.03811153396964073,
      "epoch": 4.51,
      "learning_rate": 1.4052215008870299e-05,
      "loss": 0.0526,
      "step": 4753,
      "task_loss": 0.1829942911863327
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7917244269051756,
      "compression_loss": 0.0,
      "distillation_loss": 0.08471325784921646,
      "epoch": 4.51,
      "learning_rate": 1.4042636744373225e-05,
      "loss": 0.0854,
      "step": 4754,
      "task_loss": 0.09121614694595337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.791731187993971,
      "compression_loss": 0.0,
      "distillation_loss": 0.08335433900356293,
      "epoch": 4.52,
      "learning_rate": 1.4033060470333948e-05,
      "loss": 0.0847,
      "step": 4755,
      "task_loss": 0.09658505022525787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.791737945399253,
      "compression_loss": 0.0,
      "distillation_loss": 0.09752035140991211,
      "epoch": 4.52,
      "learning_rate": 1.4023486188492052e-05,
      "loss": 0.0917,
      "step": 4756,
      "task_loss": 0.03966911882162094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7917446991220255,
      "compression_loss": 0.0,
      "distillation_loss": 0.05041617155075073,
      "epoch": 4.52,
      "learning_rate": 1.4013913900586767e-05,
      "loss": 0.0554,
      "step": 4757,
      "task_loss": 0.10072137415409088
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.791751449163292,
      "compression_loss": 0.0,
      "distillation_loss": 0.03461472690105438,
      "epoch": 4.52,
      "learning_rate": 1.4004343608356928e-05,
      "loss": 0.0325,
      "step": 4758,
      "task_loss": 0.013332528993487358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7917581955240564,
      "compression_loss": 0.0,
      "distillation_loss": 0.11117596924304962,
      "epoch": 4.52,
      "learning_rate": 1.399477531354102e-05,
      "loss": 0.1109,
      "step": 4759,
      "task_loss": 0.10825280100107193
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7917649382053221,
      "compression_loss": 0.0,
      "distillation_loss": 0.12476976960897446,
      "epoch": 4.52,
      "learning_rate": 1.3985209017877189e-05,
      "loss": 0.1206,
      "step": 4760,
      "task_loss": 0.08343881368637085
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7917716772080929,
      "compression_loss": 0.0,
      "distillation_loss": 0.10303467512130737,
      "epoch": 4.52,
      "learning_rate": 1.3975644723103185e-05,
      "loss": 0.1027,
      "step": 4761,
      "task_loss": 0.09963302314281464
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7917784125333727,
      "compression_loss": 0.0,
      "distillation_loss": 0.10136422514915466,
      "epoch": 4.52,
      "learning_rate": 1.39660824309564e-05,
      "loss": 0.0952,
      "step": 4762,
      "task_loss": 0.03952625393867493
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7917851441821648,
      "compression_loss": 0.0,
      "distillation_loss": 0.04904685169458389,
      "epoch": 4.52,
      "learning_rate": 1.3956522143173894e-05,
      "loss": 0.0537,
      "step": 4763,
      "task_loss": 0.09588026255369186
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7917918721554732,
      "compression_loss": 0.0,
      "distillation_loss": 0.04798861965537071,
      "epoch": 4.52,
      "learning_rate": 1.3946963861492319e-05,
      "loss": 0.0518,
      "step": 4764,
      "task_loss": 0.08626765012741089
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7917985964543014,
      "compression_loss": 0.0,
      "distillation_loss": 0.015419387258589268,
      "epoch": 4.53,
      "learning_rate": 1.3937407587648e-05,
      "loss": 0.0148,
      "step": 4765,
      "task_loss": 0.009636864066123962
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7918053170796532,
      "compression_loss": 0.0,
      "distillation_loss": 0.09505701810121536,
      "epoch": 4.53,
      "learning_rate": 1.3927853323376855e-05,
      "loss": 0.0958,
      "step": 4766,
      "task_loss": 0.10282032191753387
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7918120340325321,
      "compression_loss": 0.0,
      "distillation_loss": 0.07270844280719757,
      "epoch": 4.53,
      "learning_rate": 1.391830107041449e-05,
      "loss": 0.0779,
      "step": 4767,
      "task_loss": 0.12441182881593704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.791818747313942,
      "compression_loss": 0.0,
      "distillation_loss": 0.1424996554851532,
      "epoch": 4.53,
      "learning_rate": 1.390875083049611e-05,
      "loss": 0.1441,
      "step": 4768,
      "task_loss": 0.15887659788131714
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7918254569248865,
      "compression_loss": 0.0,
      "distillation_loss": 0.020403150469064713,
      "epoch": 4.53,
      "learning_rate": 1.3899202605356542e-05,
      "loss": 0.0188,
      "step": 4769,
      "task_loss": 0.00390862300992012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7918321628663691,
      "compression_loss": 0.0,
      "distillation_loss": 0.19000548124313354,
      "epoch": 4.53,
      "learning_rate": 1.38896563967303e-05,
      "loss": 0.1851,
      "step": 4770,
      "task_loss": 0.14091481268405914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7918388651393938,
      "compression_loss": 0.0,
      "distillation_loss": 0.06574898958206177,
      "epoch": 4.53,
      "learning_rate": 1.3880112206351475e-05,
      "loss": 0.0741,
      "step": 4771,
      "task_loss": 0.14917248487472534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7918455637449642,
      "compression_loss": 0.0,
      "distillation_loss": 0.21178287267684937,
      "epoch": 4.53,
      "learning_rate": 1.3870570035953811e-05,
      "loss": 0.2009,
      "step": 4772,
      "task_loss": 0.10322417318820953
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7918522586840838,
      "compression_loss": 0.0,
      "distillation_loss": 0.03703705966472626,
      "epoch": 4.53,
      "learning_rate": 1.3861029887270705e-05,
      "loss": 0.0342,
      "step": 4773,
      "task_loss": 0.008660474792122841
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7918589499577564,
      "compression_loss": 0.0,
      "distillation_loss": 0.06054367870092392,
      "epoch": 4.53,
      "learning_rate": 1.3851491762035173e-05,
      "loss": 0.0626,
      "step": 4774,
      "task_loss": 0.08063255250453949
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7918656375669858,
      "compression_loss": 0.0,
      "distillation_loss": 0.06310294568538666,
      "epoch": 4.53,
      "learning_rate": 1.3841955661979856e-05,
      "loss": 0.0585,
      "step": 4775,
      "task_loss": 0.017228560522198677
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7918723215127754,
      "compression_loss": 0.0,
      "distillation_loss": 0.05638519302010536,
      "epoch": 4.54,
      "learning_rate": 1.383242158883702e-05,
      "loss": 0.0603,
      "step": 4776,
      "task_loss": 0.09584720432758331
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7918790017961291,
      "compression_loss": 0.0,
      "distillation_loss": 0.07800333946943283,
      "epoch": 4.54,
      "learning_rate": 1.3822889544338596e-05,
      "loss": 0.0848,
      "step": 4777,
      "task_loss": 0.14598438143730164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7918856784180506,
      "compression_loss": 0.0,
      "distillation_loss": 0.03396276757121086,
      "epoch": 4.54,
      "learning_rate": 1.3813359530216113e-05,
      "loss": 0.0467,
      "step": 4778,
      "task_loss": 0.1610591858625412
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7918923513795434,
      "compression_loss": 0.0,
      "distillation_loss": 0.04296828433871269,
      "epoch": 4.54,
      "learning_rate": 1.3803831548200741e-05,
      "loss": 0.0402,
      "step": 4779,
      "task_loss": 0.014961333945393562
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7918990206816113,
      "compression_loss": 0.0,
      "distillation_loss": 0.0331374853849411,
      "epoch": 4.54,
      "learning_rate": 1.3794305600023296e-05,
      "loss": 0.0358,
      "step": 4780,
      "task_loss": 0.05958189442753792
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.791905686325258,
      "compression_loss": 0.0,
      "distillation_loss": 0.017804235219955444,
      "epoch": 4.54,
      "learning_rate": 1.37847816874142e-05,
      "loss": 0.0163,
      "step": 4781,
      "task_loss": 0.0025765616446733475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7919123483114872,
      "compression_loss": 0.0,
      "distillation_loss": 0.03256119042634964,
      "epoch": 4.54,
      "learning_rate": 1.377525981210353e-05,
      "loss": 0.0302,
      "step": 4782,
      "task_loss": 0.008887385949492455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7919190066413024,
      "compression_loss": 0.0,
      "distillation_loss": 0.020645175129175186,
      "epoch": 4.54,
      "learning_rate": 1.3765739975820962e-05,
      "loss": 0.0188,
      "step": 4783,
      "task_loss": 0.0026280879974365234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7919256613157075,
      "compression_loss": 0.0,
      "distillation_loss": 0.036273565143346786,
      "epoch": 4.54,
      "learning_rate": 1.3756222180295848e-05,
      "loss": 0.0377,
      "step": 4784,
      "task_loss": 0.05044960230588913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7919323123357062,
      "compression_loss": 0.0,
      "distillation_loss": 0.027985138818621635,
      "epoch": 4.54,
      "learning_rate": 1.3746706427257122e-05,
      "loss": 0.0387,
      "step": 4785,
      "task_loss": 0.13473522663116455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.791938959702302,
      "compression_loss": 0.0,
      "distillation_loss": 0.0791940987110138,
      "epoch": 4.55,
      "learning_rate": 1.3737192718433362e-05,
      "loss": 0.0858,
      "step": 4786,
      "task_loss": 0.1456226408481598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7919456034164986,
      "compression_loss": 0.0,
      "distillation_loss": 0.03578998148441315,
      "epoch": 4.55,
      "learning_rate": 1.3727681055552797e-05,
      "loss": 0.0327,
      "step": 4787,
      "task_loss": 0.004547275602817535
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7919522434792998,
      "compression_loss": 0.0,
      "distillation_loss": 0.044103555381298065,
      "epoch": 4.55,
      "learning_rate": 1.3718171440343253e-05,
      "loss": 0.0404,
      "step": 4788,
      "task_loss": 0.006830913946032524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7919588798917092,
      "compression_loss": 0.0,
      "distillation_loss": 0.019968142732977867,
      "epoch": 4.55,
      "learning_rate": 1.3708663874532196e-05,
      "loss": 0.0184,
      "step": 4789,
      "task_loss": 0.0039040017873048782
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7919655126547306,
      "compression_loss": 0.0,
      "distillation_loss": 0.04486335813999176,
      "epoch": 4.55,
      "learning_rate": 1.369915835984672e-05,
      "loss": 0.0521,
      "step": 4790,
      "task_loss": 0.11712378263473511
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7919721417693675,
      "compression_loss": 0.0,
      "distillation_loss": 0.07928772270679474,
      "epoch": 4.55,
      "learning_rate": 1.3689654898013568e-05,
      "loss": 0.0866,
      "step": 4791,
      "task_loss": 0.15211215615272522
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7919787672366236,
      "compression_loss": 0.0,
      "distillation_loss": 0.08449915051460266,
      "epoch": 4.55,
      "learning_rate": 1.3680153490759073e-05,
      "loss": 0.0801,
      "step": 4792,
      "task_loss": 0.04078545793890953
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7919853890575028,
      "compression_loss": 0.0,
      "distillation_loss": 0.030160902068018913,
      "epoch": 4.55,
      "learning_rate": 1.3670654139809202e-05,
      "loss": 0.0471,
      "step": 4793,
      "task_loss": 0.19971755146980286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7919920072330086,
      "compression_loss": 0.0,
      "distillation_loss": 0.07142870128154755,
      "epoch": 4.55,
      "learning_rate": 1.3661156846889584e-05,
      "loss": 0.067,
      "step": 4794,
      "task_loss": 0.027513636276125908
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7919986217641447,
      "compression_loss": 0.0,
      "distillation_loss": 0.06955306231975555,
      "epoch": 4.55,
      "learning_rate": 1.3651661613725428e-05,
      "loss": 0.0697,
      "step": 4795,
      "task_loss": 0.07086768746376038
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7920052326519148,
      "compression_loss": 0.0,
      "distillation_loss": 0.02464466169476509,
      "epoch": 4.55,
      "learning_rate": 1.3642168442041586e-05,
      "loss": 0.028,
      "step": 4796,
      "task_loss": 0.05863853171467781
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7920118398973226,
      "compression_loss": 0.0,
      "distillation_loss": 0.04839901998639107,
      "epoch": 4.56,
      "learning_rate": 1.3632677333562557e-05,
      "loss": 0.0461,
      "step": 4797,
      "task_loss": 0.0250839926302433
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7920184435013717,
      "compression_loss": 0.0,
      "distillation_loss": 0.07862424850463867,
      "epoch": 4.56,
      "learning_rate": 1.3623188290012434e-05,
      "loss": 0.0786,
      "step": 4798,
      "task_loss": 0.0784490555524826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792025043465066,
      "compression_loss": 0.0,
      "distillation_loss": 0.1141967624425888,
      "epoch": 4.56,
      "learning_rate": 1.361370131311494e-05,
      "loss": 0.1161,
      "step": 4799,
      "task_loss": 0.133285790681839
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7920316397894089,
      "compression_loss": 0.0,
      "distillation_loss": 0.027825910598039627,
      "epoch": 4.56,
      "learning_rate": 1.3604216404593442e-05,
      "loss": 0.0255,
      "step": 4800,
      "task_loss": 0.004920231178402901
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7920382324754043,
      "compression_loss": 0.0,
      "distillation_loss": 0.06682395190000534,
      "epoch": 4.56,
      "learning_rate": 1.3594733566170926e-05,
      "loss": 0.0627,
      "step": 4801,
      "task_loss": 0.025574171915650368
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7920448215240558,
      "compression_loss": 0.0,
      "distillation_loss": 0.07818278670310974,
      "epoch": 4.56,
      "learning_rate": 1.3585252799569987e-05,
      "loss": 0.0918,
      "step": 4802,
      "task_loss": 0.21402281522750854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792051406936367,
      "compression_loss": 0.0,
      "distillation_loss": 0.03720712661743164,
      "epoch": 4.56,
      "learning_rate": 1.357577410651284e-05,
      "loss": 0.0382,
      "step": 4803,
      "task_loss": 0.047226957976818085
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7920579887133417,
      "compression_loss": 0.0,
      "distillation_loss": 0.069380983710289,
      "epoch": 4.56,
      "learning_rate": 1.3566297488721352e-05,
      "loss": 0.0696,
      "step": 4804,
      "task_loss": 0.07113489508628845
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7920645668559835,
      "compression_loss": 0.0,
      "distillation_loss": 0.029062392190098763,
      "epoch": 4.56,
      "learning_rate": 1.3556822947916998e-05,
      "loss": 0.0366,
      "step": 4805,
      "task_loss": 0.10417832434177399
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7920711413652961,
      "compression_loss": 0.0,
      "distillation_loss": 0.024266686290502548,
      "epoch": 4.56,
      "learning_rate": 1.354735048582086e-05,
      "loss": 0.0301,
      "step": 4806,
      "task_loss": 0.08248142898082733
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7920777122422833,
      "compression_loss": 0.0,
      "distillation_loss": 0.06912650167942047,
      "epoch": 4.57,
      "learning_rate": 1.3537880104153644e-05,
      "loss": 0.0707,
      "step": 4807,
      "task_loss": 0.0849725604057312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7920842794879486,
      "compression_loss": 0.0,
      "distillation_loss": 0.016859427094459534,
      "epoch": 4.57,
      "learning_rate": 1.3528411804635732e-05,
      "loss": 0.0157,
      "step": 4808,
      "task_loss": 0.004863811656832695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7920908431032958,
      "compression_loss": 0.0,
      "distillation_loss": 0.023333366960287094,
      "epoch": 4.57,
      "learning_rate": 1.3518945588987062e-05,
      "loss": 0.0282,
      "step": 4809,
      "task_loss": 0.07221191376447678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7920974030893286,
      "compression_loss": 0.0,
      "distillation_loss": 0.0138099305331707,
      "epoch": 4.57,
      "learning_rate": 1.3509481458927209e-05,
      "loss": 0.0365,
      "step": 4810,
      "task_loss": 0.24029265344142914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7921039594470506,
      "compression_loss": 0.0,
      "distillation_loss": 0.0632215216755867,
      "epoch": 4.57,
      "learning_rate": 1.3500019416175396e-05,
      "loss": 0.0725,
      "step": 4811,
      "task_loss": 0.15620213747024536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7921105121774655,
      "compression_loss": 0.0,
      "distillation_loss": 0.10874255001544952,
      "epoch": 4.57,
      "learning_rate": 1.3490559462450445e-05,
      "loss": 0.103,
      "step": 4812,
      "task_loss": 0.051314111799001694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792117061281577,
      "compression_loss": 0.0,
      "distillation_loss": 0.11337900161743164,
      "epoch": 4.57,
      "learning_rate": 1.3481101599470794e-05,
      "loss": 0.106,
      "step": 4813,
      "task_loss": 0.03977866470813751
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7921236067603888,
      "compression_loss": 0.0,
      "distillation_loss": 0.14787593483924866,
      "epoch": 4.57,
      "learning_rate": 1.3471645828954504e-05,
      "loss": 0.1417,
      "step": 4814,
      "task_loss": 0.0858084037899971
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7921301486149046,
      "compression_loss": 0.0,
      "distillation_loss": 0.07990266382694244,
      "epoch": 4.57,
      "learning_rate": 1.346219215261928e-05,
      "loss": 0.0812,
      "step": 4815,
      "task_loss": 0.09328167140483856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7921366868461279,
      "compression_loss": 0.0,
      "distillation_loss": 0.029033560305833817,
      "epoch": 4.57,
      "learning_rate": 1.345274057218241e-05,
      "loss": 0.0336,
      "step": 4816,
      "task_loss": 0.07436929643154144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7921432214550627,
      "compression_loss": 0.0,
      "distillation_loss": 0.0795164555311203,
      "epoch": 4.57,
      "learning_rate": 1.3443291089360827e-05,
      "loss": 0.0783,
      "step": 4817,
      "task_loss": 0.06710808724164963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7921497524427125,
      "compression_loss": 0.0,
      "distillation_loss": 0.1971079707145691,
      "epoch": 4.58,
      "learning_rate": 1.3433843705871086e-05,
      "loss": 0.205,
      "step": 4818,
      "task_loss": 0.27602148056030273
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7921562798100809,
      "compression_loss": 0.0,
      "distillation_loss": 0.1079106554389,
      "epoch": 4.58,
      "learning_rate": 1.3424398423429335e-05,
      "loss": 0.1086,
      "step": 4819,
      "task_loss": 0.11473225057125092
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7921628035581716,
      "compression_loss": 0.0,
      "distillation_loss": 0.06394623965024948,
      "epoch": 4.58,
      "learning_rate": 1.3414955243751362e-05,
      "loss": 0.065,
      "step": 4820,
      "task_loss": 0.07409561425447464
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7921693236879885,
      "compression_loss": 0.0,
      "distillation_loss": 0.038549233227968216,
      "epoch": 4.58,
      "learning_rate": 1.3405514168552552e-05,
      "loss": 0.0359,
      "step": 4821,
      "task_loss": 0.011564519256353378
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7921758402005351,
      "compression_loss": 0.0,
      "distillation_loss": 0.035132892429828644,
      "epoch": 4.58,
      "learning_rate": 1.3396075199547944e-05,
      "loss": 0.0475,
      "step": 4822,
      "task_loss": 0.1591106653213501
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7921823530968151,
      "compression_loss": 0.0,
      "distillation_loss": 0.12006336450576782,
      "epoch": 4.58,
      "learning_rate": 1.3386638338452162e-05,
      "loss": 0.1215,
      "step": 4823,
      "task_loss": 0.13471020758152008
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7921888623778321,
      "compression_loss": 0.0,
      "distillation_loss": 0.11836139857769012,
      "epoch": 4.58,
      "learning_rate": 1.3377203586979444e-05,
      "loss": 0.1116,
      "step": 4824,
      "task_loss": 0.050610434263944626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7921953680445901,
      "compression_loss": 0.0,
      "distillation_loss": 0.03372761979699135,
      "epoch": 4.58,
      "learning_rate": 1.3367770946843671e-05,
      "loss": 0.0362,
      "step": 4825,
      "task_loss": 0.057958073914051056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922018700980924,
      "compression_loss": 0.0,
      "distillation_loss": 0.018163051456212997,
      "epoch": 4.58,
      "learning_rate": 1.3358340419758339e-05,
      "loss": 0.021,
      "step": 4826,
      "task_loss": 0.046153098344802856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792208368539343,
      "compression_loss": 0.0,
      "distillation_loss": 0.022957507520914078,
      "epoch": 4.58,
      "learning_rate": 1.3348912007436537e-05,
      "loss": 0.0211,
      "step": 4827,
      "task_loss": 0.004677103832364082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922148633693453,
      "compression_loss": 0.0,
      "distillation_loss": 0.09547284990549088,
      "epoch": 4.58,
      "learning_rate": 1.3339485711590965e-05,
      "loss": 0.0951,
      "step": 4828,
      "task_loss": 0.09193813055753708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922213545891031,
      "compression_loss": 0.0,
      "distillation_loss": 0.031036367639899254,
      "epoch": 4.59,
      "learning_rate": 1.3330061533933987e-05,
      "loss": 0.0287,
      "step": 4829,
      "task_loss": 0.008153628557920456
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922278421996202,
      "compression_loss": 0.0,
      "distillation_loss": 0.06869910657405853,
      "epoch": 4.59,
      "learning_rate": 1.3320639476177533e-05,
      "loss": 0.0683,
      "step": 4830,
      "task_loss": 0.06460073590278625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922343262019,
      "compression_loss": 0.0,
      "distillation_loss": 0.02499844878911972,
      "epoch": 4.59,
      "learning_rate": 1.3311219540033156e-05,
      "loss": 0.023,
      "step": 4831,
      "task_loss": 0.004916973412036896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922408065969465,
      "compression_loss": 0.0,
      "distillation_loss": 0.04321729391813278,
      "epoch": 4.59,
      "learning_rate": 1.3301801727212054e-05,
      "loss": 0.0431,
      "step": 4832,
      "task_loss": 0.04201708361506462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922472833857632,
      "compression_loss": 0.0,
      "distillation_loss": 0.04636396840214729,
      "epoch": 4.59,
      "learning_rate": 1.3292386039424998e-05,
      "loss": 0.0444,
      "step": 4833,
      "task_loss": 0.026649996638298035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922537565693538,
      "compression_loss": 0.0,
      "distillation_loss": 0.08710770308971405,
      "epoch": 4.59,
      "learning_rate": 1.328297247838241e-05,
      "loss": 0.0833,
      "step": 4834,
      "task_loss": 0.04921592399477959
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922602261487222,
      "compression_loss": 0.0,
      "distillation_loss": 0.05042930692434311,
      "epoch": 4.59,
      "learning_rate": 1.3273561045794294e-05,
      "loss": 0.0674,
      "step": 4835,
      "task_loss": 0.21977365016937256
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922666921248717,
      "compression_loss": 0.0,
      "distillation_loss": 0.13431815803050995,
      "epoch": 4.59,
      "learning_rate": 1.3264151743370299e-05,
      "loss": 0.1476,
      "step": 4836,
      "task_loss": 0.2666419744491577
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922731544988061,
      "compression_loss": 0.0,
      "distillation_loss": 0.10049565136432648,
      "epoch": 4.59,
      "learning_rate": 1.3254744572819658e-05,
      "loss": 0.1039,
      "step": 4837,
      "task_loss": 0.13440871238708496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922796132715293,
      "compression_loss": 0.0,
      "distillation_loss": 0.07015521079301834,
      "epoch": 4.59,
      "learning_rate": 1.324533953585122e-05,
      "loss": 0.0672,
      "step": 4838,
      "task_loss": 0.04096302390098572
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922860684440447,
      "compression_loss": 0.0,
      "distillation_loss": 0.040450118482112885,
      "epoch": 4.6,
      "learning_rate": 1.323593663417348e-05,
      "loss": 0.0478,
      "step": 4839,
      "task_loss": 0.11380739510059357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922925200173562,
      "compression_loss": 0.0,
      "distillation_loss": 0.02944648265838623,
      "epoch": 4.6,
      "learning_rate": 1.3226535869494505e-05,
      "loss": 0.0349,
      "step": 4840,
      "task_loss": 0.08415282517671585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7922989679924674,
      "compression_loss": 0.0,
      "distillation_loss": 0.08658148348331451,
      "epoch": 4.6,
      "learning_rate": 1.3217137243521981e-05,
      "loss": 0.09,
      "step": 4841,
      "task_loss": 0.12077930569648743
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792305412370382,
      "compression_loss": 0.0,
      "distillation_loss": 0.05245388299226761,
      "epoch": 4.6,
      "learning_rate": 1.3207740757963225e-05,
      "loss": 0.0538,
      "step": 4842,
      "task_loss": 0.06580542027950287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7923118531521036,
      "compression_loss": 0.0,
      "distillation_loss": 0.06145579740405083,
      "epoch": 4.6,
      "learning_rate": 1.3198346414525162e-05,
      "loss": 0.0614,
      "step": 4843,
      "task_loss": 0.06057172268629074
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792318290338636,
      "compression_loss": 0.0,
      "distillation_loss": 0.01584722474217415,
      "epoch": 4.6,
      "learning_rate": 1.318895421491431e-05,
      "loss": 0.0147,
      "step": 4844,
      "task_loss": 0.004009943455457687
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7923247239309827,
      "compression_loss": 0.0,
      "distillation_loss": 0.04391245171427727,
      "epoch": 4.6,
      "learning_rate": 1.3179564160836794e-05,
      "loss": 0.0507,
      "step": 4845,
      "task_loss": 0.11139274388551712
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7923311539301476,
      "compression_loss": 0.0,
      "distillation_loss": 0.03385882452130318,
      "epoch": 4.6,
      "learning_rate": 1.317017625399839e-05,
      "loss": 0.0316,
      "step": 4846,
      "task_loss": 0.011443352326750755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7923375803371343,
      "compression_loss": 0.0,
      "distillation_loss": 0.029054662212729454,
      "epoch": 4.6,
      "learning_rate": 1.3160790496104441e-05,
      "loss": 0.0302,
      "step": 4847,
      "task_loss": 0.040480926632881165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7923440031529465,
      "compression_loss": 0.0,
      "distillation_loss": 0.10286715626716614,
      "epoch": 4.6,
      "learning_rate": 1.3151406888859907e-05,
      "loss": 0.1156,
      "step": 4848,
      "task_loss": 0.2303019016981125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7923504223785878,
      "compression_loss": 0.0,
      "distillation_loss": 0.08016088604927063,
      "epoch": 4.6,
      "learning_rate": 1.3142025433969384e-05,
      "loss": 0.0915,
      "step": 4849,
      "task_loss": 0.19339843094348907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792356838015062,
      "compression_loss": 0.0,
      "distillation_loss": 0.03166133165359497,
      "epoch": 4.61,
      "learning_rate": 1.3132646133137053e-05,
      "loss": 0.0361,
      "step": 4850,
      "task_loss": 0.07562405616044998
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7923632500633726,
      "compression_loss": 0.0,
      "distillation_loss": 0.026462262496352196,
      "epoch": 4.61,
      "learning_rate": 1.3123268988066695e-05,
      "loss": 0.045,
      "step": 4851,
      "task_loss": 0.211809903383255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7923696585245235,
      "compression_loss": 0.0,
      "distillation_loss": 0.02861833944916725,
      "epoch": 4.61,
      "learning_rate": 1.3113894000461721e-05,
      "loss": 0.0266,
      "step": 4852,
      "task_loss": 0.007970165461301804
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7923760633995182,
      "compression_loss": 0.0,
      "distillation_loss": 0.07544773817062378,
      "epoch": 4.61,
      "learning_rate": 1.3104521172025158e-05,
      "loss": 0.0724,
      "step": 4853,
      "task_loss": 0.04472661018371582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7923824646893605,
      "compression_loss": 0.0,
      "distillation_loss": 0.06193629652261734,
      "epoch": 4.61,
      "learning_rate": 1.3095150504459614e-05,
      "loss": 0.0598,
      "step": 4854,
      "task_loss": 0.04085727035999298
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792388862395054,
      "compression_loss": 0.0,
      "distillation_loss": 0.03407544270157814,
      "epoch": 4.61,
      "learning_rate": 1.3085781999467303e-05,
      "loss": 0.044,
      "step": 4855,
      "task_loss": 0.13290613889694214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7923952565176026,
      "compression_loss": 0.0,
      "distillation_loss": 0.15552052855491638,
      "epoch": 4.61,
      "learning_rate": 1.3076415658750083e-05,
      "loss": 0.1622,
      "step": 4856,
      "task_loss": 0.2220618724822998
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7924016470580096,
      "compression_loss": 0.0,
      "distillation_loss": 0.0764533281326294,
      "epoch": 4.61,
      "learning_rate": 1.3067051484009383e-05,
      "loss": 0.0794,
      "step": 4857,
      "task_loss": 0.10618744790554047
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792408034017279,
      "compression_loss": 0.0,
      "distillation_loss": 0.08214154094457626,
      "epoch": 4.61,
      "learning_rate": 1.3057689476946238e-05,
      "loss": 0.0788,
      "step": 4858,
      "task_loss": 0.0489237904548645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7924144173964144,
      "compression_loss": 0.0,
      "distillation_loss": 0.18950366973876953,
      "epoch": 4.61,
      "learning_rate": 1.304832963926132e-05,
      "loss": 0.1862,
      "step": 4859,
      "task_loss": 0.15664705634117126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7924207971964193,
      "compression_loss": 0.0,
      "distillation_loss": 0.062167249619960785,
      "epoch": 4.62,
      "learning_rate": 1.303897197265489e-05,
      "loss": 0.0726,
      "step": 4860,
      "task_loss": 0.16604240238666534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7924271734182977,
      "compression_loss": 0.0,
      "distillation_loss": 0.02140945754945278,
      "epoch": 4.62,
      "learning_rate": 1.3029616478826805e-05,
      "loss": 0.0202,
      "step": 4861,
      "task_loss": 0.009080074727535248
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792433546063053,
      "compression_loss": 0.0,
      "distillation_loss": 0.020239079371094704,
      "epoch": 4.62,
      "learning_rate": 1.3020263159476526e-05,
      "loss": 0.0187,
      "step": 4862,
      "task_loss": 0.004914524033665657
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7924399151316892,
      "compression_loss": 0.0,
      "distillation_loss": 0.03199296444654465,
      "epoch": 4.62,
      "learning_rate": 1.301091201630315e-05,
      "loss": 0.0295,
      "step": 4863,
      "task_loss": 0.0070999301970005035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7924462806252096,
      "compression_loss": 0.0,
      "distillation_loss": 0.05161367356777191,
      "epoch": 4.62,
      "learning_rate": 1.3001563051005347e-05,
      "loss": 0.0562,
      "step": 4864,
      "task_loss": 0.09769769012928009
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7924526425446181,
      "compression_loss": 0.0,
      "distillation_loss": 0.015561849810183048,
      "epoch": 4.62,
      "learning_rate": 1.2992216265281393e-05,
      "loss": 0.0211,
      "step": 4865,
      "task_loss": 0.07106789946556091
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7924590008909185,
      "compression_loss": 0.0,
      "distillation_loss": 0.03964412212371826,
      "epoch": 4.62,
      "learning_rate": 1.2982871660829191e-05,
      "loss": 0.0413,
      "step": 4866,
      "task_loss": 0.05661403387784958
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7924653556651141,
      "compression_loss": 0.0,
      "distillation_loss": 0.030754052102565765,
      "epoch": 4.62,
      "learning_rate": 1.2973529239346227e-05,
      "loss": 0.0281,
      "step": 4867,
      "task_loss": 0.004459032788872719
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7924717068682089,
      "compression_loss": 0.0,
      "distillation_loss": 0.029556550085544586,
      "epoch": 4.62,
      "learning_rate": 1.2964189002529586e-05,
      "loss": 0.0295,
      "step": 4868,
      "task_loss": 0.028598371893167496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7924780545012066,
      "compression_loss": 0.0,
      "distillation_loss": 0.022967040538787842,
      "epoch": 4.62,
      "learning_rate": 1.2954850952075982e-05,
      "loss": 0.0212,
      "step": 4869,
      "task_loss": 0.004923565313220024
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7924843985651107,
      "compression_loss": 0.0,
      "distillation_loss": 0.07921673357486725,
      "epoch": 4.62,
      "learning_rate": 1.294551508968172e-05,
      "loss": 0.0833,
      "step": 4870,
      "task_loss": 0.11980371177196503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7924907390609249,
      "compression_loss": 0.0,
      "distillation_loss": 0.0501846969127655,
      "epoch": 4.63,
      "learning_rate": 1.2936181417042697e-05,
      "loss": 0.0469,
      "step": 4871,
      "task_loss": 0.01730802096426487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792497075989653,
      "compression_loss": 0.0,
      "distillation_loss": 0.02974601648747921,
      "epoch": 4.63,
      "learning_rate": 1.2926849935854413e-05,
      "loss": 0.0274,
      "step": 4872,
      "task_loss": 0.006090117618441582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7925034093522986,
      "compression_loss": 0.0,
      "distillation_loss": 0.02171606570482254,
      "epoch": 4.63,
      "learning_rate": 1.2917520647811987e-05,
      "loss": 0.0268,
      "step": 4873,
      "task_loss": 0.07253267616033554
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7925097391498654,
      "compression_loss": 0.0,
      "distillation_loss": 0.022985346615314484,
      "epoch": 4.63,
      "learning_rate": 1.2908193554610128e-05,
      "loss": 0.0239,
      "step": 4874,
      "task_loss": 0.03176088258624077
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7925160653833572,
      "compression_loss": 0.0,
      "distillation_loss": 0.029063984751701355,
      "epoch": 4.63,
      "learning_rate": 1.2898868657943137e-05,
      "loss": 0.0273,
      "step": 4875,
      "task_loss": 0.010947054252028465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7925223880537774,
      "compression_loss": 0.0,
      "distillation_loss": 0.03057437390089035,
      "epoch": 4.63,
      "learning_rate": 1.2889545959504939e-05,
      "loss": 0.0359,
      "step": 4876,
      "task_loss": 0.08393832296133041
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79252870716213,
      "compression_loss": 0.0,
      "distillation_loss": 0.0298734400421381,
      "epoch": 4.63,
      "learning_rate": 1.2880225460989038e-05,
      "loss": 0.0332,
      "step": 4877,
      "task_loss": 0.0632658377289772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7925350227094184,
      "compression_loss": 0.0,
      "distillation_loss": 0.0205811969935894,
      "epoch": 4.63,
      "learning_rate": 1.2870907164088557e-05,
      "loss": 0.0191,
      "step": 4878,
      "task_loss": 0.005966978147625923
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7925413346966464,
      "compression_loss": 0.0,
      "distillation_loss": 0.016166334971785545,
      "epoch": 4.63,
      "learning_rate": 1.2861591070496193e-05,
      "loss": 0.0274,
      "step": 4879,
      "task_loss": 0.1287321150302887
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7925476431248178,
      "compression_loss": 0.0,
      "distillation_loss": 0.03941744565963745,
      "epoch": 4.63,
      "learning_rate": 1.2852277181904282e-05,
      "loss": 0.0516,
      "step": 4880,
      "task_loss": 0.16162440180778503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7925539479949362,
      "compression_loss": 0.0,
      "distillation_loss": 0.06537621468305588,
      "epoch": 4.64,
      "learning_rate": 1.2842965500004728e-05,
      "loss": 0.0704,
      "step": 4881,
      "task_loss": 0.11553017050027847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7925602493080053,
      "compression_loss": 0.0,
      "distillation_loss": 0.036567624658346176,
      "epoch": 4.64,
      "learning_rate": 1.2833656026489028e-05,
      "loss": 0.034,
      "step": 4882,
      "task_loss": 0.010592048987746239
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7925665470650286,
      "compression_loss": 0.0,
      "distillation_loss": 0.022438403218984604,
      "epoch": 4.64,
      "learning_rate": 1.282434876304831e-05,
      "loss": 0.0208,
      "step": 4883,
      "task_loss": 0.006338924169540405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79257284126701,
      "compression_loss": 0.0,
      "distillation_loss": 0.07297282665967941,
      "epoch": 4.64,
      "learning_rate": 1.2815043711373285e-05,
      "loss": 0.0853,
      "step": 4884,
      "task_loss": 0.19581279158592224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7925791319149531,
      "compression_loss": 0.0,
      "distillation_loss": 0.036048613488674164,
      "epoch": 4.64,
      "learning_rate": 1.2805740873154237e-05,
      "loss": 0.0621,
      "step": 4885,
      "task_loss": 0.29630377888679504
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7925854190098616,
      "compression_loss": 0.0,
      "distillation_loss": 0.055907703936100006,
      "epoch": 4.64,
      "learning_rate": 1.2796440250081093e-05,
      "loss": 0.0643,
      "step": 4886,
      "task_loss": 0.13965827226638794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7925917025527391,
      "compression_loss": 0.0,
      "distillation_loss": 0.04197829216718674,
      "epoch": 4.64,
      "learning_rate": 1.2787141843843359e-05,
      "loss": 0.0383,
      "step": 4887,
      "task_loss": 0.0054167453199625015
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7925979825445895,
      "compression_loss": 0.0,
      "distillation_loss": 0.029558390378952026,
      "epoch": 4.64,
      "learning_rate": 1.2777845656130122e-05,
      "loss": 0.0272,
      "step": 4888,
      "task_loss": 0.006031878292560577
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7926042589864162,
      "compression_loss": 0.0,
      "distillation_loss": 0.04860628768801689,
      "epoch": 4.64,
      "learning_rate": 1.276855168863008e-05,
      "loss": 0.0644,
      "step": 4889,
      "task_loss": 0.20697267353534698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792610531879223,
      "compression_loss": 0.0,
      "distillation_loss": 0.051894620060920715,
      "epoch": 4.64,
      "learning_rate": 1.2759259943031538e-05,
      "loss": 0.057,
      "step": 4890,
      "task_loss": 0.10261577367782593
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7926168012240138,
      "compression_loss": 0.0,
      "distillation_loss": 0.027323635295033455,
      "epoch": 4.64,
      "learning_rate": 1.2749970421022381e-05,
      "loss": 0.0323,
      "step": 4891,
      "task_loss": 0.07732345908880234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792623067021792,
      "compression_loss": 0.0,
      "distillation_loss": 0.05105939507484436,
      "epoch": 4.65,
      "learning_rate": 1.2740683124290081e-05,
      "loss": 0.0573,
      "step": 4892,
      "task_loss": 0.11377457529306412
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7926293292735612,
      "compression_loss": 0.0,
      "distillation_loss": 0.014040287584066391,
      "epoch": 4.65,
      "learning_rate": 1.2731398054521748e-05,
      "loss": 0.0218,
      "step": 4893,
      "task_loss": 0.09142255038022995
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7926355879803254,
      "compression_loss": 0.0,
      "distillation_loss": 0.0386216938495636,
      "epoch": 4.65,
      "learning_rate": 1.2722115213404031e-05,
      "loss": 0.0425,
      "step": 4894,
      "task_loss": 0.07720839232206345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7926418431430882,
      "compression_loss": 0.0,
      "distillation_loss": 0.06361688673496246,
      "epoch": 4.65,
      "learning_rate": 1.2712834602623228e-05,
      "loss": 0.0605,
      "step": 4895,
      "task_loss": 0.03238803148269653
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792648094762853,
      "compression_loss": 0.0,
      "distillation_loss": 0.02098456397652626,
      "epoch": 4.65,
      "learning_rate": 1.2703556223865181e-05,
      "loss": 0.0331,
      "step": 4896,
      "task_loss": 0.14170801639556885
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7926543428406239,
      "compression_loss": 0.0,
      "distillation_loss": 0.06998811662197113,
      "epoch": 4.65,
      "learning_rate": 1.2694280078815382e-05,
      "loss": 0.0669,
      "step": 4897,
      "task_loss": 0.03889083489775658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7926605873774042,
      "compression_loss": 0.0,
      "distillation_loss": 0.03850919380784035,
      "epoch": 4.65,
      "learning_rate": 1.2685006169158869e-05,
      "loss": 0.0413,
      "step": 4898,
      "task_loss": 0.06683476269245148
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7926668283741979,
      "compression_loss": 0.0,
      "distillation_loss": 0.15333296358585358,
      "epoch": 4.65,
      "learning_rate": 1.2675734496580285e-05,
      "loss": 0.1567,
      "step": 4899,
      "task_loss": 0.18686996400356293
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7926730658320085,
      "compression_loss": 0.0,
      "distillation_loss": 0.12471291422843933,
      "epoch": 4.65,
      "learning_rate": 1.2666465062763894e-05,
      "loss": 0.1294,
      "step": 4900,
      "task_loss": 0.17144428193569183
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7926792997518396,
      "compression_loss": 0.0,
      "distillation_loss": 0.11362001299858093,
      "epoch": 4.65,
      "learning_rate": 1.2657197869393523e-05,
      "loss": 0.1113,
      "step": 4901,
      "task_loss": 0.09047464281320572
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7926855301346951,
      "compression_loss": 0.0,
      "distillation_loss": 0.029277116060256958,
      "epoch": 4.66,
      "learning_rate": 1.2647932918152606e-05,
      "loss": 0.0279,
      "step": 4902,
      "task_loss": 0.015509987249970436
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7926917569815787,
      "compression_loss": 0.0,
      "distillation_loss": 0.0541316494345665,
      "epoch": 4.66,
      "learning_rate": 1.2638670210724138e-05,
      "loss": 0.0605,
      "step": 4903,
      "task_loss": 0.11828672140836716
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7926979802934938,
      "compression_loss": 0.0,
      "distillation_loss": 0.038249123841524124,
      "epoch": 4.66,
      "learning_rate": 1.2629409748790782e-05,
      "loss": 0.0403,
      "step": 4904,
      "task_loss": 0.058266542851924896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7927042000714443,
      "compression_loss": 0.0,
      "distillation_loss": 0.055571284145116806,
      "epoch": 4.66,
      "learning_rate": 1.262015153403472e-05,
      "loss": 0.0609,
      "step": 4905,
      "task_loss": 0.10911326855421066
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7927104163164338,
      "compression_loss": 0.0,
      "distillation_loss": 0.0703677237033844,
      "epoch": 4.66,
      "learning_rate": 1.2610895568137754e-05,
      "loss": 0.0799,
      "step": 4906,
      "task_loss": 0.1653900146484375
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7927166290294662,
      "compression_loss": 0.0,
      "distillation_loss": 0.04402827471494675,
      "epoch": 4.66,
      "learning_rate": 1.2601641852781265e-05,
      "loss": 0.0499,
      "step": 4907,
      "task_loss": 0.1026904359459877
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7927228382115448,
      "compression_loss": 0.0,
      "distillation_loss": 0.0630364716053009,
      "epoch": 4.66,
      "learning_rate": 1.2592390389646258e-05,
      "loss": 0.0653,
      "step": 4908,
      "task_loss": 0.08615773171186447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7927290438636735,
      "compression_loss": 0.0,
      "distillation_loss": 0.1884252429008484,
      "epoch": 4.66,
      "learning_rate": 1.2583141180413288e-05,
      "loss": 0.1818,
      "step": 4909,
      "task_loss": 0.12265457212924957
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792735245986856,
      "compression_loss": 0.0,
      "distillation_loss": 0.030301768332719803,
      "epoch": 4.66,
      "learning_rate": 1.2573894226762518e-05,
      "loss": 0.03,
      "step": 4910,
      "task_loss": 0.02761382982134819
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7927414445820959,
      "compression_loss": 0.0,
      "distillation_loss": 0.08405353128910065,
      "epoch": 4.66,
      "learning_rate": 1.25646495303737e-05,
      "loss": 0.0817,
      "step": 4911,
      "task_loss": 0.06076852232217789
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792747639650397,
      "compression_loss": 0.0,
      "distillation_loss": 0.02579943835735321,
      "epoch": 4.66,
      "learning_rate": 1.2555407092926197e-05,
      "loss": 0.0374,
      "step": 4912,
      "task_loss": 0.14211821556091309
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7927538311927629,
      "compression_loss": 0.0,
      "distillation_loss": 0.07202491909265518,
      "epoch": 4.67,
      "learning_rate": 1.2546166916098928e-05,
      "loss": 0.0744,
      "step": 4913,
      "task_loss": 0.09626448899507523
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7927600192101972,
      "compression_loss": 0.0,
      "distillation_loss": 0.04386778920888901,
      "epoch": 4.67,
      "learning_rate": 1.253692900157041e-05,
      "loss": 0.0428,
      "step": 4914,
      "task_loss": 0.03336333483457565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7927662037037038,
      "compression_loss": 0.0,
      "distillation_loss": 0.05689205601811409,
      "epoch": 4.67,
      "learning_rate": 1.252769335101877e-05,
      "loss": 0.0683,
      "step": 4915,
      "task_loss": 0.17090968787670135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7927723846742861,
      "compression_loss": 0.0,
      "distillation_loss": 0.1544632911682129,
      "epoch": 4.67,
      "learning_rate": 1.25184599661217e-05,
      "loss": 0.1538,
      "step": 4916,
      "task_loss": 0.14771823585033417
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7927785621229481,
      "compression_loss": 0.0,
      "distillation_loss": 0.015177516266703606,
      "epoch": 4.67,
      "learning_rate": 1.2509228848556482e-05,
      "loss": 0.0142,
      "step": 4917,
      "task_loss": 0.005492331460118294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7927847360506932,
      "compression_loss": 0.0,
      "distillation_loss": 0.039150480180978775,
      "epoch": 4.67,
      "learning_rate": 1.2500000000000006e-05,
      "loss": 0.0361,
      "step": 4918,
      "task_loss": 0.00842754915356636
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7927909064585252,
      "compression_loss": 0.0,
      "distillation_loss": 0.024652540683746338,
      "epoch": 4.67,
      "learning_rate": 1.2490773422128732e-05,
      "loss": 0.0228,
      "step": 4919,
      "task_loss": 0.005637306720018387
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7927970733474479,
      "compression_loss": 0.0,
      "distillation_loss": 0.04846763238310814,
      "epoch": 4.67,
      "learning_rate": 1.2481549116618698e-05,
      "loss": 0.0448,
      "step": 4920,
      "task_loss": 0.011590449139475822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928032367184648,
      "compression_loss": 0.0,
      "distillation_loss": 0.014604704454541206,
      "epoch": 4.67,
      "learning_rate": 1.247232708514556e-05,
      "loss": 0.0136,
      "step": 4921,
      "task_loss": 0.004115687683224678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928093965725795,
      "compression_loss": 0.0,
      "distillation_loss": 0.027096299454569817,
      "epoch": 4.67,
      "learning_rate": 1.2463107329384552e-05,
      "loss": 0.0252,
      "step": 4922,
      "task_loss": 0.008570542559027672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.792815552910796,
      "compression_loss": 0.0,
      "distillation_loss": 0.023527517914772034,
      "epoch": 4.68,
      "learning_rate": 1.2453889851010473e-05,
      "loss": 0.0217,
      "step": 4923,
      "task_loss": 0.00551653653383255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928217057341178,
      "compression_loss": 0.0,
      "distillation_loss": 0.07601621747016907,
      "epoch": 4.68,
      "learning_rate": 1.2444674651697716e-05,
      "loss": 0.0799,
      "step": 4924,
      "task_loss": 0.1144181340932846
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928278550435486,
      "compression_loss": 0.0,
      "distillation_loss": 0.02135540172457695,
      "epoch": 4.68,
      "learning_rate": 1.2435461733120287e-05,
      "loss": 0.0337,
      "step": 4925,
      "task_loss": 0.1447429358959198
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928340008400919,
      "compression_loss": 0.0,
      "distillation_loss": 0.0660846009850502,
      "epoch": 4.68,
      "learning_rate": 1.2426251096951744e-05,
      "loss": 0.064,
      "step": 4926,
      "task_loss": 0.04477040097117424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928401431247517,
      "compression_loss": 0.0,
      "distillation_loss": 0.037703558802604675,
      "epoch": 4.68,
      "learning_rate": 1.2417042744865237e-05,
      "loss": 0.0531,
      "step": 4927,
      "task_loss": 0.19128406047821045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928462818985316,
      "compression_loss": 0.0,
      "distillation_loss": 0.024694714695215225,
      "epoch": 4.68,
      "learning_rate": 1.2407836678533523e-05,
      "loss": 0.0321,
      "step": 4928,
      "task_loss": 0.09900853782892227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928524171624352,
      "compression_loss": 0.0,
      "distillation_loss": 0.05749480798840523,
      "epoch": 4.68,
      "learning_rate": 1.2398632899628912e-05,
      "loss": 0.054,
      "step": 4929,
      "task_loss": 0.022426774725317955
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928585489174661,
      "compression_loss": 0.0,
      "distillation_loss": 0.029665688052773476,
      "epoch": 4.68,
      "learning_rate": 1.2389431409823336e-05,
      "loss": 0.0283,
      "step": 4930,
      "task_loss": 0.01619105413556099
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928646771646282,
      "compression_loss": 0.0,
      "distillation_loss": 0.031163331121206284,
      "epoch": 4.68,
      "learning_rate": 1.2380232210788265e-05,
      "loss": 0.0326,
      "step": 4931,
      "task_loss": 0.045856185257434845
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928708019049249,
      "compression_loss": 0.0,
      "distillation_loss": 0.08584082871675491,
      "epoch": 4.68,
      "learning_rate": 1.23710353041948e-05,
      "loss": 0.0879,
      "step": 4932,
      "task_loss": 0.10655449330806732
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928769231393602,
      "compression_loss": 0.0,
      "distillation_loss": 0.11349561810493469,
      "epoch": 4.68,
      "learning_rate": 1.2361840691713595e-05,
      "loss": 0.1096,
      "step": 4933,
      "task_loss": 0.07412466406822205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928830408689376,
      "compression_loss": 0.0,
      "distillation_loss": 0.0716467946767807,
      "epoch": 4.69,
      "learning_rate": 1.2352648375014883e-05,
      "loss": 0.0791,
      "step": 4934,
      "task_loss": 0.14578455686569214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928891550946607,
      "compression_loss": 0.0,
      "distillation_loss": 0.03932729363441467,
      "epoch": 4.69,
      "learning_rate": 1.2343458355768513e-05,
      "loss": 0.0449,
      "step": 4935,
      "task_loss": 0.09523576498031616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7928952658175334,
      "compression_loss": 0.0,
      "distillation_loss": 0.06750982254743576,
      "epoch": 4.69,
      "learning_rate": 1.233427063564389e-05,
      "loss": 0.0789,
      "step": 4936,
      "task_loss": 0.18135161697864532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929013730385593,
      "compression_loss": 0.0,
      "distillation_loss": 0.05587318539619446,
      "epoch": 4.69,
      "learning_rate": 1.2325085216309994e-05,
      "loss": 0.066,
      "step": 4937,
      "task_loss": 0.156914621591568
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929074767587421,
      "compression_loss": 0.0,
      "distillation_loss": 0.044295355677604675,
      "epoch": 4.69,
      "learning_rate": 1.231590209943541e-05,
      "loss": 0.0426,
      "step": 4938,
      "task_loss": 0.02747635915875435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929135769790853,
      "compression_loss": 0.0,
      "distillation_loss": 0.010653965175151825,
      "epoch": 4.69,
      "learning_rate": 1.230672128668831e-05,
      "loss": 0.0163,
      "step": 4939,
      "task_loss": 0.06707858294248581
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929196737005928,
      "compression_loss": 0.0,
      "distillation_loss": 0.10607406497001648,
      "epoch": 4.69,
      "learning_rate": 1.2297542779736417e-05,
      "loss": 0.1061,
      "step": 4940,
      "task_loss": 0.10609476268291473
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929257669242682,
      "compression_loss": 0.0,
      "distillation_loss": 0.025449592620134354,
      "epoch": 4.69,
      "learning_rate": 1.2288366580247047e-05,
      "loss": 0.0324,
      "step": 4941,
      "task_loss": 0.09515959024429321
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929318566511152,
      "compression_loss": 0.0,
      "distillation_loss": 0.042621444910764694,
      "epoch": 4.69,
      "learning_rate": 1.2279192689887115e-05,
      "loss": 0.0455,
      "step": 4942,
      "task_loss": 0.07145626842975616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929379428821375,
      "compression_loss": 0.0,
      "distillation_loss": 0.01993633061647415,
      "epoch": 4.69,
      "learning_rate": 1.2270021110323096e-05,
      "loss": 0.0186,
      "step": 4943,
      "task_loss": 0.00634993240237236
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929440256183387,
      "compression_loss": 0.0,
      "distillation_loss": 0.014678630977869034,
      "epoch": 4.7,
      "learning_rate": 1.2260851843221039e-05,
      "loss": 0.0236,
      "step": 4944,
      "task_loss": 0.10428041964769363
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929501048607225,
      "compression_loss": 0.0,
      "distillation_loss": 0.02224035933613777,
      "epoch": 4.7,
      "learning_rate": 1.225168489024661e-05,
      "loss": 0.0263,
      "step": 4945,
      "task_loss": 0.0633140280842781
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929561806102927,
      "compression_loss": 0.0,
      "distillation_loss": 0.0771671012043953,
      "epoch": 4.7,
      "learning_rate": 1.2242520253065004e-05,
      "loss": 0.0854,
      "step": 4946,
      "task_loss": 0.15969133377075195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929622528680528,
      "compression_loss": 0.0,
      "distillation_loss": 0.012984257191419601,
      "epoch": 4.7,
      "learning_rate": 1.2233357933341047e-05,
      "loss": 0.0249,
      "step": 4947,
      "task_loss": 0.13174743950366974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929683216350066,
      "compression_loss": 0.0,
      "distillation_loss": 0.013885335996747017,
      "epoch": 4.7,
      "learning_rate": 1.2224197932739096e-05,
      "loss": 0.0128,
      "step": 4948,
      "task_loss": 0.0031508170068264008
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929743869121578,
      "compression_loss": 0.0,
      "distillation_loss": 0.01262105256319046,
      "epoch": 4.7,
      "learning_rate": 1.2215040252923127e-05,
      "loss": 0.021,
      "step": 4949,
      "task_loss": 0.09617089480161667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929804487005101,
      "compression_loss": 0.0,
      "distillation_loss": 0.07396592199802399,
      "epoch": 4.7,
      "learning_rate": 1.2205884895556672e-05,
      "loss": 0.0722,
      "step": 4950,
      "task_loss": 0.05672897398471832
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929865070010671,
      "compression_loss": 0.0,
      "distillation_loss": 0.019163163378834724,
      "epoch": 4.7,
      "learning_rate": 1.219673186230283e-05,
      "loss": 0.018,
      "step": 4951,
      "task_loss": 0.007804272696375847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7929925618148325,
      "compression_loss": 0.0,
      "distillation_loss": 0.06364789605140686,
      "epoch": 4.7,
      "learning_rate": 1.2187581154824317e-05,
      "loss": 0.0665,
      "step": 4952,
      "task_loss": 0.09249377250671387
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79299861314281,
      "compression_loss": 0.0,
      "distillation_loss": 0.09652909636497498,
      "epoch": 4.7,
      "learning_rate": 1.2178432774783394e-05,
      "loss": 0.0985,
      "step": 4953,
      "task_loss": 0.11579639464616776
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930046609860032,
      "compression_loss": 0.0,
      "distillation_loss": 0.0369391068816185,
      "epoch": 4.7,
      "learning_rate": 1.2169286723841897e-05,
      "loss": 0.0342,
      "step": 4954,
      "task_loss": 0.009912891313433647
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930107053454158,
      "compression_loss": 0.0,
      "distillation_loss": 0.03373286873102188,
      "epoch": 4.71,
      "learning_rate": 1.216014300366126e-05,
      "loss": 0.0389,
      "step": 4955,
      "task_loss": 0.08519351482391357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930167462220518,
      "compression_loss": 0.0,
      "distillation_loss": 0.024067046120762825,
      "epoch": 4.71,
      "learning_rate": 1.2151001615902493e-05,
      "loss": 0.032,
      "step": 4956,
      "task_loss": 0.10371088981628418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930227836169144,
      "compression_loss": 0.0,
      "distillation_loss": 0.14827989041805267,
      "epoch": 4.71,
      "learning_rate": 1.2141862562226165e-05,
      "loss": 0.1495,
      "step": 4957,
      "task_loss": 0.15999506413936615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930288175310075,
      "compression_loss": 0.0,
      "distillation_loss": 0.04894055798649788,
      "epoch": 4.71,
      "learning_rate": 1.2132725844292416e-05,
      "loss": 0.0561,
      "step": 4958,
      "task_loss": 0.12008975446224213
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930348479653349,
      "compression_loss": 0.0,
      "distillation_loss": 0.08717606216669083,
      "epoch": 4.71,
      "learning_rate": 1.2123591463760997e-05,
      "loss": 0.0892,
      "step": 4959,
      "task_loss": 0.10745367407798767
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930408749209001,
      "compression_loss": 0.0,
      "distillation_loss": 0.030964989215135574,
      "epoch": 4.71,
      "learning_rate": 1.2114459422291205e-05,
      "loss": 0.0285,
      "step": 4960,
      "task_loss": 0.006043644621968269
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930468983987069,
      "compression_loss": 0.0,
      "distillation_loss": 0.021385207772254944,
      "epoch": 4.71,
      "learning_rate": 1.2105329721541903e-05,
      "loss": 0.03,
      "step": 4961,
      "task_loss": 0.10744792968034744
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930529183997588,
      "compression_loss": 0.0,
      "distillation_loss": 0.1010364517569542,
      "epoch": 4.71,
      "learning_rate": 1.2096202363171571e-05,
      "loss": 0.0969,
      "step": 4962,
      "task_loss": 0.05933345481753349
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930589349250597,
      "compression_loss": 0.0,
      "distillation_loss": 0.0767788365483284,
      "epoch": 4.71,
      "learning_rate": 1.2087077348838214e-05,
      "loss": 0.0806,
      "step": 4963,
      "task_loss": 0.11461080610752106
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930649479756132,
      "compression_loss": 0.0,
      "distillation_loss": 0.12463347613811493,
      "epoch": 4.71,
      "learning_rate": 1.2077954680199455e-05,
      "loss": 0.1252,
      "step": 4964,
      "task_loss": 0.12991327047348022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793070957552423,
      "compression_loss": 0.0,
      "distillation_loss": 0.02803153544664383,
      "epoch": 4.72,
      "learning_rate": 1.2068834358912454e-05,
      "loss": 0.0265,
      "step": 4965,
      "task_loss": 0.012602541595697403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930769636564927,
      "compression_loss": 0.0,
      "distillation_loss": 0.036852188408374786,
      "epoch": 4.72,
      "learning_rate": 1.2059716386633977e-05,
      "loss": 0.0409,
      "step": 4966,
      "task_loss": 0.07686673104763031
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930829662888261,
      "compression_loss": 0.0,
      "distillation_loss": 0.01830023154616356,
      "epoch": 4.72,
      "learning_rate": 1.205060076502034e-05,
      "loss": 0.0173,
      "step": 4967,
      "task_loss": 0.00809447094798088
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930889654504267,
      "compression_loss": 0.0,
      "distillation_loss": 0.22687265276908875,
      "epoch": 4.72,
      "learning_rate": 1.204148749572743e-05,
      "loss": 0.2179,
      "step": 4968,
      "task_loss": 0.136878103017807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7930949611422985,
      "compression_loss": 0.0,
      "distillation_loss": 0.030822109431028366,
      "epoch": 4.72,
      "learning_rate": 1.2032376580410731e-05,
      "loss": 0.0291,
      "step": 4969,
      "task_loss": 0.013280034065246582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931009533654448,
      "compression_loss": 0.0,
      "distillation_loss": 0.016060050576925278,
      "epoch": 4.72,
      "learning_rate": 1.2023268020725285e-05,
      "loss": 0.0427,
      "step": 4970,
      "task_loss": 0.2822893261909485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931069421208696,
      "compression_loss": 0.0,
      "distillation_loss": 0.16523435711860657,
      "epoch": 4.72,
      "learning_rate": 1.2014161818325687e-05,
      "loss": 0.1601,
      "step": 4971,
      "task_loss": 0.11406517028808594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931129274095764,
      "compression_loss": 0.0,
      "distillation_loss": 0.04404463618993759,
      "epoch": 4.72,
      "learning_rate": 1.2005057974866135e-05,
      "loss": 0.047,
      "step": 4972,
      "task_loss": 0.07401682436466217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931189092325689,
      "compression_loss": 0.0,
      "distillation_loss": 0.03614489734172821,
      "epoch": 4.72,
      "learning_rate": 1.1995956492000397e-05,
      "loss": 0.0331,
      "step": 4973,
      "task_loss": 0.005481433123350143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931248875908509,
      "compression_loss": 0.0,
      "distillation_loss": 0.048779651522636414,
      "epoch": 4.72,
      "learning_rate": 1.1986857371381788e-05,
      "loss": 0.0466,
      "step": 4974,
      "task_loss": 0.027400122955441475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931308624854259,
      "compression_loss": 0.0,
      "distillation_loss": 0.03478461876511574,
      "epoch": 4.72,
      "learning_rate": 1.19777606146632e-05,
      "loss": 0.0362,
      "step": 4975,
      "task_loss": 0.04934518039226532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931368339172977,
      "compression_loss": 0.0,
      "distillation_loss": 0.10953269898891449,
      "epoch": 4.73,
      "learning_rate": 1.1968666223497124e-05,
      "loss": 0.1125,
      "step": 4976,
      "task_loss": 0.1394570916891098
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79314280188747,
      "compression_loss": 0.0,
      "distillation_loss": 0.039646346122026443,
      "epoch": 4.73,
      "learning_rate": 1.1959574199535586e-05,
      "loss": 0.0363,
      "step": 4977,
      "task_loss": 0.005907488986849785
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931487663969464,
      "compression_loss": 0.0,
      "distillation_loss": 0.017676934599876404,
      "epoch": 4.73,
      "learning_rate": 1.1950484544430191e-05,
      "loss": 0.0241,
      "step": 4978,
      "task_loss": 0.08169616758823395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931547274467307,
      "compression_loss": 0.0,
      "distillation_loss": 0.035570040345191956,
      "epoch": 4.73,
      "learning_rate": 1.1941397259832133e-05,
      "loss": 0.0371,
      "step": 4979,
      "task_loss": 0.05074803903698921
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931606850378264,
      "compression_loss": 0.0,
      "distillation_loss": 0.058902185410261154,
      "epoch": 4.73,
      "learning_rate": 1.1932312347392154e-05,
      "loss": 0.0549,
      "step": 4980,
      "task_loss": 0.019008172675967216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931666391712374,
      "compression_loss": 0.0,
      "distillation_loss": 0.014614107087254524,
      "epoch": 4.73,
      "learning_rate": 1.1923229808760564e-05,
      "loss": 0.0136,
      "step": 4981,
      "task_loss": 0.0042775776237249374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931725898479672,
      "compression_loss": 0.0,
      "distillation_loss": 0.16022302210330963,
      "epoch": 4.73,
      "learning_rate": 1.1914149645587256e-05,
      "loss": 0.1527,
      "step": 4982,
      "task_loss": 0.08545684814453125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931785370690195,
      "compression_loss": 0.0,
      "distillation_loss": 0.03780351206660271,
      "epoch": 4.73,
      "learning_rate": 1.1905071859521697e-05,
      "loss": 0.0506,
      "step": 4983,
      "task_loss": 0.16610009968280792
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931844808353982,
      "compression_loss": 0.0,
      "distillation_loss": 0.04032333940267563,
      "epoch": 4.73,
      "learning_rate": 1.1895996452212898e-05,
      "loss": 0.0395,
      "step": 4984,
      "task_loss": 0.032581619918346405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931904211481067,
      "compression_loss": 0.0,
      "distillation_loss": 0.03943778946995735,
      "epoch": 4.73,
      "learning_rate": 1.1886923425309445e-05,
      "loss": 0.0375,
      "step": 4985,
      "task_loss": 0.020389195531606674
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7931963580081488,
      "compression_loss": 0.0,
      "distillation_loss": 0.0650925487279892,
      "epoch": 4.74,
      "learning_rate": 1.1877852780459518e-05,
      "loss": 0.0637,
      "step": 4986,
      "task_loss": 0.05133982002735138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932022914165282,
      "compression_loss": 0.0,
      "distillation_loss": 0.023483579978346825,
      "epoch": 4.74,
      "learning_rate": 1.1868784519310826e-05,
      "loss": 0.023,
      "step": 4987,
      "task_loss": 0.01817641593515873
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932082213742486,
      "compression_loss": 0.0,
      "distillation_loss": 0.038998499512672424,
      "epoch": 4.74,
      "learning_rate": 1.185971864351067e-05,
      "loss": 0.0598,
      "step": 4988,
      "task_loss": 0.24691206216812134
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932141478823136,
      "compression_loss": 0.0,
      "distillation_loss": 0.08431239426136017,
      "epoch": 4.74,
      "learning_rate": 1.1850655154705886e-05,
      "loss": 0.078,
      "step": 4989,
      "task_loss": 0.020718101412057877
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793220070941727,
      "compression_loss": 0.0,
      "distillation_loss": 0.01440352015197277,
      "epoch": 4.74,
      "learning_rate": 1.184159405454294e-05,
      "loss": 0.0134,
      "step": 4990,
      "task_loss": 0.004175456240773201
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932259905534923,
      "compression_loss": 0.0,
      "distillation_loss": 0.039678964763879776,
      "epoch": 4.74,
      "learning_rate": 1.1832535344667806e-05,
      "loss": 0.0511,
      "step": 4991,
      "task_loss": 0.1535629779100418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932319067186134,
      "compression_loss": 0.0,
      "distillation_loss": 0.187263622879982,
      "epoch": 4.74,
      "learning_rate": 1.1823479026726031e-05,
      "loss": 0.1935,
      "step": 4992,
      "task_loss": 0.24967440962791443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932378194380938,
      "compression_loss": 0.0,
      "distillation_loss": 0.013006833381950855,
      "epoch": 4.74,
      "learning_rate": 1.1814425102362761e-05,
      "loss": 0.0231,
      "step": 4993,
      "task_loss": 0.11390747874975204
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932437287129374,
      "compression_loss": 0.0,
      "distillation_loss": 0.03500215336680412,
      "epoch": 4.74,
      "learning_rate": 1.1805373573222673e-05,
      "loss": 0.0377,
      "step": 4994,
      "task_loss": 0.06191583350300789
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932496345441475,
      "compression_loss": 0.0,
      "distillation_loss": 0.036350950598716736,
      "epoch": 4.74,
      "learning_rate": 1.1796324440950021e-05,
      "loss": 0.0333,
      "step": 4995,
      "task_loss": 0.0061857327818870544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932555369327282,
      "compression_loss": 0.0,
      "distillation_loss": 0.053805820643901825,
      "epoch": 4.74,
      "learning_rate": 1.1787277707188616e-05,
      "loss": 0.0542,
      "step": 4996,
      "task_loss": 0.058030955493450165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932614358796829,
      "compression_loss": 0.0,
      "distillation_loss": 0.12426917254924774,
      "epoch": 4.75,
      "learning_rate": 1.1778233373581857e-05,
      "loss": 0.119,
      "step": 4997,
      "task_loss": 0.07191702723503113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932673313860155,
      "compression_loss": 0.0,
      "distillation_loss": 0.026698037981987,
      "epoch": 4.75,
      "learning_rate": 1.1769191441772672e-05,
      "loss": 0.0249,
      "step": 4998,
      "task_loss": 0.008508618921041489
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932732234527294,
      "compression_loss": 0.0,
      "distillation_loss": 0.02993636205792427,
      "epoch": 4.75,
      "learning_rate": 1.1760151913403583e-05,
      "loss": 0.0303,
      "step": 4999,
      "task_loss": 0.03406871110200882
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932791120808286,
      "compression_loss": 0.0,
      "distillation_loss": 0.11071500182151794,
      "epoch": 4.75,
      "learning_rate": 1.1751114790116672e-05,
      "loss": 0.1129,
      "step": 5000,
      "task_loss": 0.13242202997207642
    },
    {
      "epoch": 4.75,
      "eval_accuracy": 0.8967889908256881,
      "eval_loss": 0.4209362268447876,
      "eval_runtime": 17.9495,
      "eval_samples_per_second": 48.581,
      "eval_steps_per_second": 6.073,
      "step": 5000
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932849972713165,
      "compression_loss": 0.0,
      "distillation_loss": 0.022589676082134247,
      "epoch": 4.75,
      "learning_rate": 1.1742080073553565e-05,
      "loss": 0.0305,
      "step": 5001,
      "task_loss": 0.10146450996398926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793290879025197,
      "compression_loss": 0.0,
      "distillation_loss": 0.13711608946323395,
      "epoch": 4.75,
      "learning_rate": 1.1733047765355466e-05,
      "loss": 0.1323,
      "step": 5002,
      "task_loss": 0.08903352916240692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7932967573434737,
      "compression_loss": 0.0,
      "distillation_loss": 0.06641732901334763,
      "epoch": 4.75,
      "learning_rate": 1.1724017867163125e-05,
      "loss": 0.0615,
      "step": 5003,
      "task_loss": 0.01706705428659916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933026322271504,
      "compression_loss": 0.0,
      "distillation_loss": 0.10634138435125351,
      "epoch": 4.75,
      "learning_rate": 1.1714990380616884e-05,
      "loss": 0.1025,
      "step": 5004,
      "task_loss": 0.06803934276103973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933085036772304,
      "compression_loss": 0.0,
      "distillation_loss": 0.032394587993621826,
      "epoch": 4.75,
      "learning_rate": 1.1705965307356624e-05,
      "loss": 0.0461,
      "step": 5005,
      "task_loss": 0.16980022192001343
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933143716947177,
      "compression_loss": 0.0,
      "distillation_loss": 0.03210689127445221,
      "epoch": 4.75,
      "learning_rate": 1.169694264902178e-05,
      "loss": 0.0516,
      "step": 5006,
      "task_loss": 0.22744594514369965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933202362806161,
      "compression_loss": 0.0,
      "distillation_loss": 0.05615242198109627,
      "epoch": 4.75,
      "learning_rate": 1.168792240725137e-05,
      "loss": 0.0532,
      "step": 5007,
      "task_loss": 0.0271142590790987
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793326097435929,
      "compression_loss": 0.0,
      "distillation_loss": 0.182695671916008,
      "epoch": 4.76,
      "learning_rate": 1.1678904583683979e-05,
      "loss": 0.1831,
      "step": 5008,
      "task_loss": 0.18713723123073578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933319551616602,
      "compression_loss": 0.0,
      "distillation_loss": 0.07359650731086731,
      "epoch": 4.76,
      "learning_rate": 1.1669889179957725e-05,
      "loss": 0.0844,
      "step": 5009,
      "task_loss": 0.1816483587026596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933378094588133,
      "compression_loss": 0.0,
      "distillation_loss": 0.11011994630098343,
      "epoch": 4.76,
      "learning_rate": 1.1660876197710288e-05,
      "loss": 0.1121,
      "step": 5010,
      "task_loss": 0.13008421659469604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933436603283921,
      "compression_loss": 0.0,
      "distillation_loss": 0.02712424099445343,
      "epoch": 4.76,
      "learning_rate": 1.1651865638578944e-05,
      "loss": 0.0252,
      "step": 5011,
      "task_loss": 0.008356472477316856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933495077714002,
      "compression_loss": 0.0,
      "distillation_loss": 0.058398209512233734,
      "epoch": 4.76,
      "learning_rate": 1.1642857504200491e-05,
      "loss": 0.053,
      "step": 5012,
      "task_loss": 0.004138745367527008
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933553517888414,
      "compression_loss": 0.0,
      "distillation_loss": 0.02487853169441223,
      "epoch": 4.76,
      "learning_rate": 1.1633851796211292e-05,
      "loss": 0.0327,
      "step": 5013,
      "task_loss": 0.10304756462574005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933611923817193,
      "compression_loss": 0.0,
      "distillation_loss": 0.09818457067012787,
      "epoch": 4.76,
      "learning_rate": 1.1624848516247295e-05,
      "loss": 0.0971,
      "step": 5014,
      "task_loss": 0.08704036474227905
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933670295510376,
      "compression_loss": 0.0,
      "distillation_loss": 0.08643116801977158,
      "epoch": 4.76,
      "learning_rate": 1.1615847665943971e-05,
      "loss": 0.0919,
      "step": 5015,
      "task_loss": 0.14145499467849731
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933728632977998,
      "compression_loss": 0.0,
      "distillation_loss": 0.027642257511615753,
      "epoch": 4.76,
      "learning_rate": 1.1606849246936389e-05,
      "loss": 0.0295,
      "step": 5016,
      "task_loss": 0.045945487916469574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933786936230098,
      "compression_loss": 0.0,
      "distillation_loss": 0.024577468633651733,
      "epoch": 4.76,
      "learning_rate": 1.1597853260859128e-05,
      "loss": 0.026,
      "step": 5017,
      "task_loss": 0.039100244641304016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933845205276713,
      "compression_loss": 0.0,
      "distillation_loss": 0.0331391841173172,
      "epoch": 4.77,
      "learning_rate": 1.1588859709346384e-05,
      "loss": 0.0595,
      "step": 5018,
      "task_loss": 0.29628539085388184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793390344012788,
      "compression_loss": 0.0,
      "distillation_loss": 0.03754423186182976,
      "epoch": 4.77,
      "learning_rate": 1.157986859403186e-05,
      "loss": 0.0382,
      "step": 5019,
      "task_loss": 0.04380743205547333
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7933961640793633,
      "compression_loss": 0.0,
      "distillation_loss": 0.11798399686813354,
      "epoch": 4.77,
      "learning_rate": 1.1570879916548827e-05,
      "loss": 0.1147,
      "step": 5020,
      "task_loss": 0.0851517766714096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934019807284012,
      "compression_loss": 0.0,
      "distillation_loss": 0.034330565482378006,
      "epoch": 4.77,
      "learning_rate": 1.1561893678530141e-05,
      "loss": 0.0472,
      "step": 5021,
      "task_loss": 0.16320303082466125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934077939609051,
      "compression_loss": 0.0,
      "distillation_loss": 0.04643068462610245,
      "epoch": 4.77,
      "learning_rate": 1.155290988160819e-05,
      "loss": 0.0523,
      "step": 5022,
      "task_loss": 0.10536113381385803
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793413603777879,
      "compression_loss": 0.0,
      "distillation_loss": 0.022104818373918533,
      "epoch": 4.77,
      "learning_rate": 1.154392852741491e-05,
      "loss": 0.036,
      "step": 5023,
      "task_loss": 0.1612529158592224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934194101803264,
      "compression_loss": 0.0,
      "distillation_loss": 0.09352529048919678,
      "epoch": 4.77,
      "learning_rate": 1.153494961758182e-05,
      "loss": 0.106,
      "step": 5024,
      "task_loss": 0.21867401897907257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934252131692509,
      "compression_loss": 0.0,
      "distillation_loss": 0.02358633652329445,
      "epoch": 4.77,
      "learning_rate": 1.1525973153739989e-05,
      "loss": 0.0252,
      "step": 5025,
      "task_loss": 0.03958255797624588
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934310127456563,
      "compression_loss": 0.0,
      "distillation_loss": 0.04170753434300423,
      "epoch": 4.77,
      "learning_rate": 1.1516999137520023e-05,
      "loss": 0.0451,
      "step": 5026,
      "task_loss": 0.07600220292806625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934368089105464,
      "compression_loss": 0.0,
      "distillation_loss": 0.047912318259477615,
      "epoch": 4.77,
      "learning_rate": 1.1508027570552094e-05,
      "loss": 0.0595,
      "step": 5027,
      "task_loss": 0.16383595764636993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934426016649245,
      "compression_loss": 0.0,
      "distillation_loss": 0.06721566617488861,
      "epoch": 4.77,
      "learning_rate": 1.1499058454465941e-05,
      "loss": 0.0778,
      "step": 5028,
      "task_loss": 0.1731707900762558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934483910097948,
      "compression_loss": 0.0,
      "distillation_loss": 0.04140312969684601,
      "epoch": 4.78,
      "learning_rate": 1.1490091790890842e-05,
      "loss": 0.0483,
      "step": 5029,
      "task_loss": 0.11057362705469131
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934541769461606,
      "compression_loss": 0.0,
      "distillation_loss": 0.04200680926442146,
      "epoch": 4.78,
      "learning_rate": 1.1481127581455626e-05,
      "loss": 0.0499,
      "step": 5030,
      "task_loss": 0.12135404348373413
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934599594750257,
      "compression_loss": 0.0,
      "distillation_loss": 0.04865646734833717,
      "epoch": 4.78,
      "learning_rate": 1.1472165827788697e-05,
      "loss": 0.0534,
      "step": 5031,
      "task_loss": 0.09640198945999146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934657385973938,
      "compression_loss": 0.0,
      "distillation_loss": 0.08559805154800415,
      "epoch": 4.78,
      "learning_rate": 1.1463206531518001e-05,
      "loss": 0.0895,
      "step": 5032,
      "task_loss": 0.12415318191051483
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934715143142684,
      "compression_loss": 0.0,
      "distillation_loss": 0.04011882096529007,
      "epoch": 4.78,
      "learning_rate": 1.1454249694271021e-05,
      "loss": 0.0579,
      "step": 5033,
      "task_loss": 0.21819335222244263
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934772866266535,
      "compression_loss": 0.0,
      "distillation_loss": 0.08635386824607849,
      "epoch": 4.78,
      "learning_rate": 1.144529531767482e-05,
      "loss": 0.09,
      "step": 5034,
      "task_loss": 0.12273044884204865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934830555355527,
      "compression_loss": 0.0,
      "distillation_loss": 0.02714727818965912,
      "epoch": 4.78,
      "learning_rate": 1.1436343403356017e-05,
      "loss": 0.0251,
      "step": 5035,
      "task_loss": 0.006460065022110939
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934888210419694,
      "compression_loss": 0.0,
      "distillation_loss": 0.080211341381073,
      "epoch": 4.78,
      "learning_rate": 1.1427393952940754e-05,
      "loss": 0.0851,
      "step": 5036,
      "task_loss": 0.12943881750106812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7934945831469076,
      "compression_loss": 0.0,
      "distillation_loss": 0.04694737493991852,
      "epoch": 4.78,
      "learning_rate": 1.1418446968054741e-05,
      "loss": 0.0637,
      "step": 5037,
      "task_loss": 0.21490800380706787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935003418513709,
      "compression_loss": 0.0,
      "distillation_loss": 0.1910262107849121,
      "epoch": 4.78,
      "learning_rate": 1.1409502450323254e-05,
      "loss": 0.1973,
      "step": 5038,
      "task_loss": 0.25344181060791016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935060971563629,
      "compression_loss": 0.0,
      "distillation_loss": 0.08850134164094925,
      "epoch": 4.79,
      "learning_rate": 1.1400560401371097e-05,
      "loss": 0.0971,
      "step": 5039,
      "task_loss": 0.17405250668525696
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935118490628873,
      "compression_loss": 0.0,
      "distillation_loss": 0.023128552362322807,
      "epoch": 4.79,
      "learning_rate": 1.1391620822822629e-05,
      "loss": 0.0234,
      "step": 5040,
      "task_loss": 0.025910664349794388
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935175975719478,
      "compression_loss": 0.0,
      "distillation_loss": 0.02330312877893448,
      "epoch": 4.79,
      "learning_rate": 1.1382683716301781e-05,
      "loss": 0.0262,
      "step": 5041,
      "task_loss": 0.0519491508603096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935233426845482,
      "compression_loss": 0.0,
      "distillation_loss": 0.029029665514826775,
      "epoch": 4.79,
      "learning_rate": 1.1373749083432025e-05,
      "loss": 0.0293,
      "step": 5042,
      "task_loss": 0.03185213357210159
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793529084401692,
      "compression_loss": 0.0,
      "distillation_loss": 0.030532198026776314,
      "epoch": 4.79,
      "learning_rate": 1.1364816925836372e-05,
      "loss": 0.0405,
      "step": 5043,
      "task_loss": 0.13049139082431793
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793534822724383,
      "compression_loss": 0.0,
      "distillation_loss": 0.030790645629167557,
      "epoch": 4.79,
      "learning_rate": 1.1355887245137383e-05,
      "loss": 0.0355,
      "step": 5044,
      "task_loss": 0.07748115062713623
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935405576536249,
      "compression_loss": 0.0,
      "distillation_loss": 0.10416917502880096,
      "epoch": 4.79,
      "learning_rate": 1.1346960042957197e-05,
      "loss": 0.1008,
      "step": 5045,
      "task_loss": 0.07013058662414551
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935462891904211,
      "compression_loss": 0.0,
      "distillation_loss": 0.018903907388448715,
      "epoch": 4.79,
      "learning_rate": 1.1338035320917473e-05,
      "loss": 0.0232,
      "step": 5046,
      "task_loss": 0.06144241988658905
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935520173357757,
      "compression_loss": 0.0,
      "distillation_loss": 0.012658301740884781,
      "epoch": 4.79,
      "learning_rate": 1.1329113080639419e-05,
      "loss": 0.029,
      "step": 5047,
      "task_loss": 0.17617405951023102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935577420906922,
      "compression_loss": 0.0,
      "distillation_loss": 0.025603320449590683,
      "epoch": 4.79,
      "learning_rate": 1.1320193323743824e-05,
      "loss": 0.026,
      "step": 5048,
      "task_loss": 0.02996073290705681
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935634634561742,
      "compression_loss": 0.0,
      "distillation_loss": 0.04193894937634468,
      "epoch": 4.79,
      "learning_rate": 1.1311276051850994e-05,
      "loss": 0.0599,
      "step": 5049,
      "task_loss": 0.22172468900680542
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935691814332255,
      "compression_loss": 0.0,
      "distillation_loss": 0.089395672082901,
      "epoch": 4.8,
      "learning_rate": 1.1302361266580786e-05,
      "loss": 0.0867,
      "step": 5050,
      "task_loss": 0.0627622902393341
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935748960228497,
      "compression_loss": 0.0,
      "distillation_loss": 0.03509131073951721,
      "epoch": 4.8,
      "learning_rate": 1.1293448969552623e-05,
      "loss": 0.0442,
      "step": 5051,
      "task_loss": 0.12643598020076752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935806072260505,
      "compression_loss": 0.0,
      "distillation_loss": 0.035416148602962494,
      "epoch": 4.8,
      "learning_rate": 1.1284539162385474e-05,
      "loss": 0.0432,
      "step": 5052,
      "task_loss": 0.1130717545747757
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935863150438317,
      "compression_loss": 0.0,
      "distillation_loss": 0.038104936480522156,
      "epoch": 4.8,
      "learning_rate": 1.127563184669784e-05,
      "loss": 0.0569,
      "step": 5053,
      "task_loss": 0.226011723279953
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935920194771968,
      "compression_loss": 0.0,
      "distillation_loss": 0.07715682685375214,
      "epoch": 4.8,
      "learning_rate": 1.1266727024107771e-05,
      "loss": 0.0907,
      "step": 5054,
      "task_loss": 0.2121979296207428
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7935977205271496,
      "compression_loss": 0.0,
      "distillation_loss": 0.02819431945681572,
      "epoch": 4.8,
      "learning_rate": 1.1257824696232888e-05,
      "loss": 0.0261,
      "step": 5055,
      "task_loss": 0.007308483123779297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7936034181946937,
      "compression_loss": 0.0,
      "distillation_loss": 0.020847158506512642,
      "epoch": 4.8,
      "learning_rate": 1.124892486469033e-05,
      "loss": 0.0193,
      "step": 5056,
      "task_loss": 0.005755210295319557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7936091124808329,
      "compression_loss": 0.0,
      "distillation_loss": 0.03235051780939102,
      "epoch": 4.8,
      "learning_rate": 1.1240027531096786e-05,
      "loss": 0.0299,
      "step": 5057,
      "task_loss": 0.007758015766739845
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7936148033865708,
      "compression_loss": 0.0,
      "distillation_loss": 0.19748008251190186,
      "epoch": 4.8,
      "learning_rate": 1.1231132697068523e-05,
      "loss": 0.1925,
      "step": 5058,
      "task_loss": 0.14814594388008118
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793620490912911,
      "compression_loss": 0.0,
      "distillation_loss": 0.013191369362175465,
      "epoch": 4.8,
      "learning_rate": 1.1222240364221303e-05,
      "loss": 0.0211,
      "step": 5059,
      "task_loss": 0.09272238612174988
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7936261750608573,
      "compression_loss": 0.0,
      "distillation_loss": 0.05488280951976776,
      "epoch": 4.81,
      "learning_rate": 1.1213350534170488e-05,
      "loss": 0.0512,
      "step": 5060,
      "task_loss": 0.018499650061130524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7936318558314134,
      "compression_loss": 0.0,
      "distillation_loss": 0.025873934850096703,
      "epoch": 4.81,
      "learning_rate": 1.1204463208530936e-05,
      "loss": 0.0244,
      "step": 5061,
      "task_loss": 0.01143595576286316
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793637533225583,
      "compression_loss": 0.0,
      "distillation_loss": 0.05775732174515724,
      "epoch": 4.81,
      "learning_rate": 1.1195578388917092e-05,
      "loss": 0.0579,
      "step": 5062,
      "task_loss": 0.05883469432592392
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7936432072443695,
      "compression_loss": 0.0,
      "distillation_loss": 0.058215852826833725,
      "epoch": 4.81,
      "learning_rate": 1.1186696076942916e-05,
      "loss": 0.0542,
      "step": 5063,
      "task_loss": 0.018394893035292625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793648877888777,
      "compression_loss": 0.0,
      "distillation_loss": 0.07067988067865372,
      "epoch": 4.81,
      "learning_rate": 1.1177816274221911e-05,
      "loss": 0.0808,
      "step": 5064,
      "task_loss": 0.17217287421226501
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7936545451598089,
      "compression_loss": 0.0,
      "distillation_loss": 0.03770400956273079,
      "epoch": 4.81,
      "learning_rate": 1.116893898236716e-05,
      "loss": 0.0355,
      "step": 5065,
      "task_loss": 0.015920959413051605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793660209058469,
      "compression_loss": 0.0,
      "distillation_loss": 0.049285177141427994,
      "epoch": 4.81,
      "learning_rate": 1.1160064202991254e-05,
      "loss": 0.0594,
      "step": 5066,
      "task_loss": 0.14998885989189148
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793665869585761,
      "compression_loss": 0.0,
      "distillation_loss": 0.03805050998926163,
      "epoch": 4.81,
      "learning_rate": 1.115119193770633e-05,
      "loss": 0.0408,
      "step": 5067,
      "task_loss": 0.06544838100671768
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7936715267426885,
      "compression_loss": 0.0,
      "distillation_loss": 0.021483074873685837,
      "epoch": 4.81,
      "learning_rate": 1.1142322188124088e-05,
      "loss": 0.0442,
      "step": 5068,
      "task_loss": 0.24901482462882996
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7936771805302552,
      "compression_loss": 0.0,
      "distillation_loss": 0.11341410130262375,
      "epoch": 4.81,
      "learning_rate": 1.1133454955855768e-05,
      "loss": 0.1099,
      "step": 5069,
      "task_loss": 0.0778903216123581
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7936828309494648,
      "compression_loss": 0.0,
      "distillation_loss": 0.1032809242606163,
      "epoch": 4.81,
      "learning_rate": 1.1124590242512137e-05,
      "loss": 0.0993,
      "step": 5070,
      "task_loss": 0.06375472992658615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793688478001321,
      "compression_loss": 0.0,
      "distillation_loss": 0.04365667700767517,
      "epoch": 4.82,
      "learning_rate": 1.1115728049703503e-05,
      "loss": 0.0504,
      "step": 5071,
      "task_loss": 0.11084698140621185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7936941216868274,
      "compression_loss": 0.0,
      "distillation_loss": 0.015161161310970783,
      "epoch": 4.82,
      "learning_rate": 1.1106868379039747e-05,
      "loss": 0.0141,
      "step": 5072,
      "task_loss": 0.004572039470076561
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7936997620069878,
      "compression_loss": 0.0,
      "distillation_loss": 0.038352642208337784,
      "epoch": 4.82,
      "learning_rate": 1.109801123213026e-05,
      "loss": 0.0374,
      "step": 5073,
      "task_loss": 0.02898401767015457
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937053989628058,
      "compression_loss": 0.0,
      "distillation_loss": 0.07865383476018906,
      "epoch": 4.82,
      "learning_rate": 1.1089156610583985e-05,
      "loss": 0.0769,
      "step": 5074,
      "task_loss": 0.061387475579977036
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937110325552851,
      "compression_loss": 0.0,
      "distillation_loss": 0.024030856788158417,
      "epoch": 4.82,
      "learning_rate": 1.1080304516009399e-05,
      "loss": 0.024,
      "step": 5075,
      "task_loss": 0.023738959804177284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937166627854294,
      "compression_loss": 0.0,
      "distillation_loss": 0.0827215313911438,
      "epoch": 4.82,
      "learning_rate": 1.1071454950014533e-05,
      "loss": 0.1135,
      "step": 5076,
      "task_loss": 0.3905654847621918
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937222896542424,
      "compression_loss": 0.0,
      "distillation_loss": 0.06419490277767181,
      "epoch": 4.82,
      "learning_rate": 1.1062607914206965e-05,
      "loss": 0.071,
      "step": 5077,
      "task_loss": 0.13255758583545685
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937279131627277,
      "compression_loss": 0.0,
      "distillation_loss": 0.044310204684734344,
      "epoch": 4.82,
      "learning_rate": 1.1053763410193788e-05,
      "loss": 0.057,
      "step": 5078,
      "task_loss": 0.17076875269412994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793733533311889,
      "compression_loss": 0.0,
      "distillation_loss": 0.05891291797161102,
      "epoch": 4.82,
      "learning_rate": 1.104492143958166e-05,
      "loss": 0.0659,
      "step": 5079,
      "task_loss": 0.1289033591747284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937391501027301,
      "compression_loss": 0.0,
      "distillation_loss": 0.05876559019088745,
      "epoch": 4.82,
      "learning_rate": 1.1036082003976759e-05,
      "loss": 0.0594,
      "step": 5080,
      "task_loss": 0.06525548547506332
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937447635362546,
      "compression_loss": 0.0,
      "distillation_loss": 0.1539030373096466,
      "epoch": 4.83,
      "learning_rate": 1.1027245104984816e-05,
      "loss": 0.1464,
      "step": 5081,
      "task_loss": 0.07844631373882294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937503736134661,
      "compression_loss": 0.0,
      "distillation_loss": 0.12115428596735,
      "epoch": 4.83,
      "learning_rate": 1.1018410744211081e-05,
      "loss": 0.1157,
      "step": 5082,
      "task_loss": 0.06699676811695099
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937559803353684,
      "compression_loss": 0.0,
      "distillation_loss": 0.03977685421705246,
      "epoch": 4.83,
      "learning_rate": 1.100957892326038e-05,
      "loss": 0.0385,
      "step": 5083,
      "task_loss": 0.027261190116405487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937615837029652,
      "compression_loss": 0.0,
      "distillation_loss": 0.1250099539756775,
      "epoch": 4.83,
      "learning_rate": 1.100074964373705e-05,
      "loss": 0.1183,
      "step": 5084,
      "task_loss": 0.058270517736673355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937671837172601,
      "compression_loss": 0.0,
      "distillation_loss": 0.015824340283870697,
      "epoch": 4.83,
      "learning_rate": 1.0991922907244944e-05,
      "loss": 0.0147,
      "step": 5085,
      "task_loss": 0.004903359338641167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937727803792568,
      "compression_loss": 0.0,
      "distillation_loss": 0.01482314057648182,
      "epoch": 4.83,
      "learning_rate": 1.0983098715387526e-05,
      "loss": 0.0138,
      "step": 5086,
      "task_loss": 0.004439596086740494
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793778373689959,
      "compression_loss": 0.0,
      "distillation_loss": 0.038871392607688904,
      "epoch": 4.83,
      "learning_rate": 1.0974277069767735e-05,
      "loss": 0.044,
      "step": 5087,
      "task_loss": 0.09021726995706558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937839636503703,
      "compression_loss": 0.0,
      "distillation_loss": 0.0341789647936821,
      "epoch": 4.83,
      "learning_rate": 1.096545797198806e-05,
      "loss": 0.045,
      "step": 5088,
      "task_loss": 0.14275771379470825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937895502614946,
      "compression_loss": 0.0,
      "distillation_loss": 0.02304195612668991,
      "epoch": 4.83,
      "learning_rate": 1.095664142365053e-05,
      "loss": 0.0342,
      "step": 5089,
      "task_loss": 0.13422459363937378
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7937951335243353,
      "compression_loss": 0.0,
      "distillation_loss": 0.027511442080140114,
      "epoch": 4.83,
      "learning_rate": 1.0947827426356725e-05,
      "loss": 0.0257,
      "step": 5090,
      "task_loss": 0.009677274152636528
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938007134398962,
      "compression_loss": 0.0,
      "distillation_loss": 0.039502017199993134,
      "epoch": 4.83,
      "learning_rate": 1.0939015981707746e-05,
      "loss": 0.054,
      "step": 5091,
      "task_loss": 0.18460404872894287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938062900091811,
      "compression_loss": 0.0,
      "distillation_loss": 0.03704231232404709,
      "epoch": 4.84,
      "learning_rate": 1.093020709130422e-05,
      "loss": 0.0372,
      "step": 5092,
      "task_loss": 0.038229990750551224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938118632331935,
      "compression_loss": 0.0,
      "distillation_loss": 0.05441391468048096,
      "epoch": 4.84,
      "learning_rate": 1.0921400756746338e-05,
      "loss": 0.0582,
      "step": 5093,
      "task_loss": 0.09189236164093018
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938174331129373,
      "compression_loss": 0.0,
      "distillation_loss": 0.1751764416694641,
      "epoch": 4.84,
      "learning_rate": 1.0912596979633824e-05,
      "loss": 0.173,
      "step": 5094,
      "task_loss": 0.15349790453910828
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938229996494159,
      "compression_loss": 0.0,
      "distillation_loss": 0.03634584695100784,
      "epoch": 4.84,
      "learning_rate": 1.0903795761565914e-05,
      "loss": 0.0338,
      "step": 5095,
      "task_loss": 0.010769536718726158
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938285628436332,
      "compression_loss": 0.0,
      "distillation_loss": 0.06358032673597336,
      "epoch": 4.84,
      "learning_rate": 1.0894997104141378e-05,
      "loss": 0.0612,
      "step": 5096,
      "task_loss": 0.03997098281979561
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938341226965929,
      "compression_loss": 0.0,
      "distillation_loss": 0.06305442750453949,
      "epoch": 4.84,
      "learning_rate": 1.088620100895856e-05,
      "loss": 0.0676,
      "step": 5097,
      "task_loss": 0.10872484743595123
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938396792092985,
      "compression_loss": 0.0,
      "distillation_loss": 0.02636984921991825,
      "epoch": 4.84,
      "learning_rate": 1.0877407477615304e-05,
      "loss": 0.025,
      "step": 5098,
      "task_loss": 0.013119250535964966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938452323827537,
      "compression_loss": 0.0,
      "distillation_loss": 0.04463554173707962,
      "epoch": 4.84,
      "learning_rate": 1.0868616511708982e-05,
      "loss": 0.0485,
      "step": 5099,
      "task_loss": 0.0837230458855629
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938507822179625,
      "compression_loss": 0.0,
      "distillation_loss": 0.045190710574388504,
      "epoch": 4.84,
      "learning_rate": 1.085982811283654e-05,
      "loss": 0.0637,
      "step": 5100,
      "task_loss": 0.23051756620407104
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938563287159282,
      "compression_loss": 0.0,
      "distillation_loss": 0.05444833263754845,
      "epoch": 4.84,
      "learning_rate": 1.085104228259442e-05,
      "loss": 0.0613,
      "step": 5101,
      "task_loss": 0.12304200232028961
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938618718776546,
      "compression_loss": 0.0,
      "distillation_loss": 0.03785646706819534,
      "epoch": 4.85,
      "learning_rate": 1.0842259022578602e-05,
      "loss": 0.0397,
      "step": 5102,
      "task_loss": 0.05668614059686661
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938674117041453,
      "compression_loss": 0.0,
      "distillation_loss": 0.04682941734790802,
      "epoch": 4.85,
      "learning_rate": 1.0833478334384617e-05,
      "loss": 0.0546,
      "step": 5103,
      "task_loss": 0.12421214580535889
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938729481964042,
      "compression_loss": 0.0,
      "distillation_loss": 0.09900999814271927,
      "epoch": 4.85,
      "learning_rate": 1.082470021960753e-05,
      "loss": 0.0984,
      "step": 5104,
      "task_loss": 0.09260520339012146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938784813554349,
      "compression_loss": 0.0,
      "distillation_loss": 0.021003127098083496,
      "epoch": 4.85,
      "learning_rate": 1.0815924679841916e-05,
      "loss": 0.0387,
      "step": 5105,
      "task_loss": 0.19845835864543915
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938840111822411,
      "compression_loss": 0.0,
      "distillation_loss": 0.038200899958610535,
      "epoch": 4.85,
      "learning_rate": 1.0807151716681885e-05,
      "loss": 0.0421,
      "step": 5106,
      "task_loss": 0.0776185691356659
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938895376778262,
      "compression_loss": 0.0,
      "distillation_loss": 0.016974829137325287,
      "epoch": 4.85,
      "learning_rate": 1.0798381331721109e-05,
      "loss": 0.03,
      "step": 5107,
      "task_loss": 0.14757771790027618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7938950608431943,
      "compression_loss": 0.0,
      "distillation_loss": 0.038715705275535583,
      "epoch": 4.85,
      "learning_rate": 1.0789613526552758e-05,
      "loss": 0.0371,
      "step": 5108,
      "task_loss": 0.02297482080757618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939005806793489,
      "compression_loss": 0.0,
      "distillation_loss": 0.029004612937569618,
      "epoch": 4.85,
      "learning_rate": 1.0780848302769542e-05,
      "loss": 0.036,
      "step": 5109,
      "task_loss": 0.09945765137672424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939060971872937,
      "compression_loss": 0.0,
      "distillation_loss": 0.018345588818192482,
      "epoch": 4.85,
      "learning_rate": 1.0772085661963708e-05,
      "loss": 0.0224,
      "step": 5110,
      "task_loss": 0.05867563933134079
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939116103680323,
      "compression_loss": 0.0,
      "distillation_loss": 0.026313165202736855,
      "epoch": 4.85,
      "learning_rate": 1.0763325605727048e-05,
      "loss": 0.024,
      "step": 5111,
      "task_loss": 0.003129279240965843
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939171202225684,
      "compression_loss": 0.0,
      "distillation_loss": 0.0785897821187973,
      "epoch": 4.85,
      "learning_rate": 1.0754568135650856e-05,
      "loss": 0.0924,
      "step": 5112,
      "task_loss": 0.2167954295873642
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939226267519057,
      "compression_loss": 0.0,
      "distillation_loss": 0.028556518256664276,
      "epoch": 4.86,
      "learning_rate": 1.0745813253325957e-05,
      "loss": 0.0266,
      "step": 5113,
      "task_loss": 0.0085418950766325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939281299570481,
      "compression_loss": 0.0,
      "distillation_loss": 0.020152829587459564,
      "epoch": 4.86,
      "learning_rate": 1.073706096034274e-05,
      "loss": 0.0343,
      "step": 5114,
      "task_loss": 0.161162868142128
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793933629838999,
      "compression_loss": 0.0,
      "distillation_loss": 0.03302818909287453,
      "epoch": 4.86,
      "learning_rate": 1.072831125829109e-05,
      "loss": 0.0397,
      "step": 5115,
      "task_loss": 0.09927168488502502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939391263987621,
      "compression_loss": 0.0,
      "distillation_loss": 0.027194611728191376,
      "epoch": 4.86,
      "learning_rate": 1.0719564148760427e-05,
      "loss": 0.0256,
      "step": 5116,
      "task_loss": 0.011119823902845383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939446196373412,
      "compression_loss": 0.0,
      "distillation_loss": 0.03556310385465622,
      "epoch": 4.86,
      "learning_rate": 1.0710819633339719e-05,
      "loss": 0.0339,
      "step": 5117,
      "task_loss": 0.018816489726305008
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939501095557399,
      "compression_loss": 0.0,
      "distillation_loss": 0.3605664372444153,
      "epoch": 4.86,
      "learning_rate": 1.0702077713617442e-05,
      "loss": 0.3543,
      "step": 5118,
      "task_loss": 0.2975212633609772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939555961549619,
      "compression_loss": 0.0,
      "distillation_loss": 0.02403501234948635,
      "epoch": 4.86,
      "learning_rate": 1.0693338391181598e-05,
      "loss": 0.0224,
      "step": 5119,
      "task_loss": 0.007670925930142403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793961079436011,
      "compression_loss": 0.0,
      "distillation_loss": 0.13072296977043152,
      "epoch": 4.86,
      "learning_rate": 1.0684601667619736e-05,
      "loss": 0.1322,
      "step": 5120,
      "task_loss": 0.1455729752779007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939665593998908,
      "compression_loss": 0.0,
      "distillation_loss": 0.03021019697189331,
      "epoch": 4.86,
      "learning_rate": 1.0675867544518935e-05,
      "loss": 0.0285,
      "step": 5121,
      "task_loss": 0.01261189952492714
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939720360476049,
      "compression_loss": 0.0,
      "distillation_loss": 0.09756823629140854,
      "epoch": 4.86,
      "learning_rate": 1.0667136023465782e-05,
      "loss": 0.0941,
      "step": 5122,
      "task_loss": 0.06310638785362244
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.793977509380157,
      "compression_loss": 0.0,
      "distillation_loss": 0.020023830235004425,
      "epoch": 4.87,
      "learning_rate": 1.0658407106046389e-05,
      "loss": 0.0209,
      "step": 5123,
      "task_loss": 0.029044657945632935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939829793985509,
      "compression_loss": 0.0,
      "distillation_loss": 0.06327167898416519,
      "epoch": 4.87,
      "learning_rate": 1.0649680793846423e-05,
      "loss": 0.0785,
      "step": 5124,
      "task_loss": 0.21557585895061493
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939884461037902,
      "compression_loss": 0.0,
      "distillation_loss": 0.0167409535497427,
      "epoch": 4.87,
      "learning_rate": 1.0640957088451054e-05,
      "loss": 0.0207,
      "step": 5125,
      "task_loss": 0.056279633194208145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939939094968785,
      "compression_loss": 0.0,
      "distillation_loss": 0.03627139329910278,
      "epoch": 4.87,
      "learning_rate": 1.0632235991444972e-05,
      "loss": 0.0342,
      "step": 5126,
      "task_loss": 0.01587931625545025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7939993695788197,
      "compression_loss": 0.0,
      "distillation_loss": 0.023063641041517258,
      "epoch": 4.87,
      "learning_rate": 1.0623517504412429e-05,
      "loss": 0.0296,
      "step": 5127,
      "task_loss": 0.08890461921691895
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940048263506173,
      "compression_loss": 0.0,
      "distillation_loss": 0.021907242015004158,
      "epoch": 4.87,
      "learning_rate": 1.061480162893716e-05,
      "loss": 0.0297,
      "step": 5128,
      "task_loss": 0.09995594620704651
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940102798132751,
      "compression_loss": 0.0,
      "distillation_loss": 0.05496376007795334,
      "epoch": 4.87,
      "learning_rate": 1.0606088366602462e-05,
      "loss": 0.0606,
      "step": 5129,
      "task_loss": 0.11096987873315811
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940157299677967,
      "compression_loss": 0.0,
      "distillation_loss": 0.04084121435880661,
      "epoch": 4.87,
      "learning_rate": 1.0597377718991127e-05,
      "loss": 0.0394,
      "step": 5130,
      "task_loss": 0.02599833346903324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940211768151857,
      "compression_loss": 0.0,
      "distillation_loss": 0.08682038635015488,
      "epoch": 4.87,
      "learning_rate": 1.0588669687685502e-05,
      "loss": 0.0918,
      "step": 5131,
      "task_loss": 0.13667771220207214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794026620356446,
      "compression_loss": 0.0,
      "distillation_loss": 0.056324686855077744,
      "epoch": 4.87,
      "learning_rate": 1.0579964274267434e-05,
      "loss": 0.0604,
      "step": 5132,
      "task_loss": 0.0968530923128128
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940320605925811,
      "compression_loss": 0.0,
      "distillation_loss": 0.022295914590358734,
      "epoch": 4.87,
      "learning_rate": 1.0571261480318295e-05,
      "loss": 0.0266,
      "step": 5133,
      "task_loss": 0.0648619681596756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940374975245948,
      "compression_loss": 0.0,
      "distillation_loss": 0.019165407866239548,
      "epoch": 4.88,
      "learning_rate": 1.0562561307419005e-05,
      "loss": 0.0193,
      "step": 5134,
      "task_loss": 0.020621633157134056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940429311534907,
      "compression_loss": 0.0,
      "distillation_loss": 0.10466399788856506,
      "epoch": 4.88,
      "learning_rate": 1.0553863757149985e-05,
      "loss": 0.1121,
      "step": 5135,
      "task_loss": 0.17872130870819092
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940483614802726,
      "compression_loss": 0.0,
      "distillation_loss": 0.0792841836810112,
      "epoch": 4.88,
      "learning_rate": 1.0545168831091178e-05,
      "loss": 0.0759,
      "step": 5136,
      "task_loss": 0.04564621299505234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794053788505944,
      "compression_loss": 0.0,
      "distillation_loss": 0.038627780973911285,
      "epoch": 4.88,
      "learning_rate": 1.0536476530822071e-05,
      "loss": 0.0547,
      "step": 5137,
      "task_loss": 0.19965079426765442
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940592122315087,
      "compression_loss": 0.0,
      "distillation_loss": 0.030025865882635117,
      "epoch": 4.88,
      "learning_rate": 1.0527786857921665e-05,
      "loss": 0.0306,
      "step": 5138,
      "task_loss": 0.03600824251770973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940646326579703,
      "compression_loss": 0.0,
      "distillation_loss": 0.054387472569942474,
      "epoch": 4.88,
      "learning_rate": 1.0519099813968475e-05,
      "loss": 0.0584,
      "step": 5139,
      "task_loss": 0.0948576033115387
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940700497863327,
      "compression_loss": 0.0,
      "distillation_loss": 0.04773625731468201,
      "epoch": 4.88,
      "learning_rate": 1.0510415400540538e-05,
      "loss": 0.0494,
      "step": 5140,
      "task_loss": 0.06462717801332474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940754636175993,
      "compression_loss": 0.0,
      "distillation_loss": 0.06973198801279068,
      "epoch": 4.88,
      "learning_rate": 1.0501733619215432e-05,
      "loss": 0.0686,
      "step": 5141,
      "task_loss": 0.05870828405022621
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940808741527738,
      "compression_loss": 0.0,
      "distillation_loss": 0.08635716885328293,
      "epoch": 4.88,
      "learning_rate": 1.049305447157024e-05,
      "loss": 0.1117,
      "step": 5142,
      "task_loss": 0.34026485681533813
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940862813928602,
      "compression_loss": 0.0,
      "distillation_loss": 0.019729439169168472,
      "epoch": 4.88,
      "learning_rate": 1.048437795918156e-05,
      "loss": 0.0181,
      "step": 5143,
      "task_loss": 0.0036925822496414185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940916853388619,
      "compression_loss": 0.0,
      "distillation_loss": 0.03241331875324249,
      "epoch": 4.89,
      "learning_rate": 1.0475704083625543e-05,
      "loss": 0.0368,
      "step": 5144,
      "task_loss": 0.07607822120189667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7940970859917826,
      "compression_loss": 0.0,
      "distillation_loss": 0.033501721918582916,
      "epoch": 4.89,
      "learning_rate": 1.0467032846477818e-05,
      "loss": 0.0338,
      "step": 5145,
      "task_loss": 0.03676898777484894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941024833526261,
      "compression_loss": 0.0,
      "distillation_loss": 0.04217281937599182,
      "epoch": 4.89,
      "learning_rate": 1.0458364249313577e-05,
      "loss": 0.0449,
      "step": 5146,
      "task_loss": 0.06914474070072174
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941078774223961,
      "compression_loss": 0.0,
      "distillation_loss": 0.06948964297771454,
      "epoch": 4.89,
      "learning_rate": 1.0449698293707494e-05,
      "loss": 0.0717,
      "step": 5147,
      "task_loss": 0.09196187555789948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794113268202096,
      "compression_loss": 0.0,
      "distillation_loss": 0.16506507992744446,
      "epoch": 4.89,
      "learning_rate": 1.0441034981233802e-05,
      "loss": 0.1604,
      "step": 5148,
      "task_loss": 0.11810018122196198
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941186556927298,
      "compression_loss": 0.0,
      "distillation_loss": 0.05315421149134636,
      "epoch": 4.89,
      "learning_rate": 1.043237431346622e-05,
      "loss": 0.0599,
      "step": 5149,
      "task_loss": 0.12041810899972916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794124039895301,
      "compression_loss": 0.0,
      "distillation_loss": 0.1707717776298523,
      "epoch": 4.89,
      "learning_rate": 1.0423716291977995e-05,
      "loss": 0.1788,
      "step": 5150,
      "task_loss": 0.2513020634651184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941294208108134,
      "compression_loss": 0.0,
      "distillation_loss": 0.08805613964796066,
      "epoch": 4.89,
      "learning_rate": 1.0415060918341915e-05,
      "loss": 0.0987,
      "step": 5151,
      "task_loss": 0.1942642778158188
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941347984402707,
      "compression_loss": 0.0,
      "distillation_loss": 0.017222080379724503,
      "epoch": 4.89,
      "learning_rate": 1.0406408194130259e-05,
      "loss": 0.026,
      "step": 5152,
      "task_loss": 0.10548414289951324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941401727846764,
      "compression_loss": 0.0,
      "distillation_loss": 0.15333212912082672,
      "epoch": 4.89,
      "learning_rate": 1.0397758120914838e-05,
      "loss": 0.1523,
      "step": 5153,
      "task_loss": 0.1433287113904953
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941455438450343,
      "compression_loss": 0.0,
      "distillation_loss": 0.036811452358961105,
      "epoch": 4.89,
      "learning_rate": 1.0389110700266965e-05,
      "loss": 0.0339,
      "step": 5154,
      "task_loss": 0.007728196680545807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941509116223481,
      "compression_loss": 0.0,
      "distillation_loss": 0.02540665678679943,
      "epoch": 4.9,
      "learning_rate": 1.0380465933757516e-05,
      "loss": 0.0312,
      "step": 5155,
      "task_loss": 0.08360700309276581
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941562761176214,
      "compression_loss": 0.0,
      "distillation_loss": 0.036480970680713654,
      "epoch": 4.9,
      "learning_rate": 1.037182382295684e-05,
      "loss": 0.0338,
      "step": 5156,
      "task_loss": 0.009625611826777458
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794161637331858,
      "compression_loss": 0.0,
      "distillation_loss": 0.16506238281726837,
      "epoch": 4.9,
      "learning_rate": 1.0363184369434803e-05,
      "loss": 0.1557,
      "step": 5157,
      "task_loss": 0.07142955809831619
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941669952660615,
      "compression_loss": 0.0,
      "distillation_loss": 0.13784891366958618,
      "epoch": 4.9,
      "learning_rate": 1.0354547574760828e-05,
      "loss": 0.1407,
      "step": 5158,
      "task_loss": 0.16680915653705597
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941723499212356,
      "compression_loss": 0.0,
      "distillation_loss": 0.05448282137513161,
      "epoch": 4.9,
      "learning_rate": 1.034591344050382e-05,
      "loss": 0.0615,
      "step": 5159,
      "task_loss": 0.125069722533226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941777012983839,
      "compression_loss": 0.0,
      "distillation_loss": 0.0830635204911232,
      "epoch": 4.9,
      "learning_rate": 1.0337281968232213e-05,
      "loss": 0.0912,
      "step": 5160,
      "task_loss": 0.1648905873298645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941830493985104,
      "compression_loss": 0.0,
      "distillation_loss": 0.09780248999595642,
      "epoch": 4.9,
      "learning_rate": 1.032865315951394e-05,
      "loss": 0.1198,
      "step": 5161,
      "task_loss": 0.3177647292613983
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941883942226183,
      "compression_loss": 0.0,
      "distillation_loss": 0.030118845403194427,
      "epoch": 4.9,
      "learning_rate": 1.0320027015916481e-05,
      "loss": 0.0312,
      "step": 5162,
      "task_loss": 0.04085350036621094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7941937357717117,
      "compression_loss": 0.0,
      "distillation_loss": 0.019481580704450607,
      "epoch": 4.9,
      "learning_rate": 1.0311403539006823e-05,
      "loss": 0.0233,
      "step": 5163,
      "task_loss": 0.05726565048098564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794199074046794,
      "compression_loss": 0.0,
      "distillation_loss": 0.08498305082321167,
      "epoch": 4.9,
      "learning_rate": 1.0302782730351443e-05,
      "loss": 0.0813,
      "step": 5164,
      "task_loss": 0.047981880605220795
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794204409048869,
      "compression_loss": 0.0,
      "distillation_loss": 0.09883970767259598,
      "epoch": 4.91,
      "learning_rate": 1.0294164591516372e-05,
      "loss": 0.1042,
      "step": 5165,
      "task_loss": 0.15241563320159912
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942097407789405,
      "compression_loss": 0.0,
      "distillation_loss": 0.13842526078224182,
      "epoch": 4.91,
      "learning_rate": 1.0285549124067126e-05,
      "loss": 0.1384,
      "step": 5166,
      "task_loss": 0.13855035603046417
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794215069238012,
      "compression_loss": 0.0,
      "distillation_loss": 0.2038910686969757,
      "epoch": 4.91,
      "learning_rate": 1.0276936329568748e-05,
      "loss": 0.2098,
      "step": 5167,
      "task_loss": 0.26268768310546875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942203944270871,
      "compression_loss": 0.0,
      "distillation_loss": 0.09594913572072983,
      "epoch": 4.91,
      "learning_rate": 1.0268326209585784e-05,
      "loss": 0.0974,
      "step": 5168,
      "task_loss": 0.11042402684688568
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942257163471699,
      "compression_loss": 0.0,
      "distillation_loss": 0.09930403530597687,
      "epoch": 4.91,
      "learning_rate": 1.0259718765682321e-05,
      "loss": 0.1025,
      "step": 5169,
      "task_loss": 0.13175342977046967
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942310349992636,
      "compression_loss": 0.0,
      "distillation_loss": 0.19986680150032043,
      "epoch": 4.91,
      "learning_rate": 1.0251113999421935e-05,
      "loss": 0.1976,
      "step": 5170,
      "task_loss": 0.1776677966117859
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942363503843721,
      "compression_loss": 0.0,
      "distillation_loss": 0.07311776280403137,
      "epoch": 4.91,
      "learning_rate": 1.0242511912367706e-05,
      "loss": 0.0678,
      "step": 5171,
      "task_loss": 0.019507795572280884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942416625034991,
      "compression_loss": 0.0,
      "distillation_loss": 0.041407689452171326,
      "epoch": 4.91,
      "learning_rate": 1.0233912506082277e-05,
      "loss": 0.0407,
      "step": 5172,
      "task_loss": 0.03435212001204491
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942469713576482,
      "compression_loss": 0.0,
      "distillation_loss": 0.03155471384525299,
      "epoch": 4.91,
      "learning_rate": 1.0225315782127756e-05,
      "loss": 0.0368,
      "step": 5173,
      "task_loss": 0.0840642973780632
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942522769478233,
      "compression_loss": 0.0,
      "distillation_loss": 0.059278104454278946,
      "epoch": 4.91,
      "learning_rate": 1.0216721742065777e-05,
      "loss": 0.0573,
      "step": 5174,
      "task_loss": 0.03901209309697151
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942575792750277,
      "compression_loss": 0.0,
      "distillation_loss": 0.09744984656572342,
      "epoch": 4.91,
      "learning_rate": 1.0208130387457485e-05,
      "loss": 0.1025,
      "step": 5175,
      "task_loss": 0.14773344993591309
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942628783402654,
      "compression_loss": 0.0,
      "distillation_loss": 0.04598250985145569,
      "epoch": 4.92,
      "learning_rate": 1.0199541719863554e-05,
      "loss": 0.0494,
      "step": 5176,
      "task_loss": 0.08029758185148239
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79426817414454,
      "compression_loss": 0.0,
      "distillation_loss": 0.06423640996217728,
      "epoch": 4.92,
      "learning_rate": 1.0190955740844147e-05,
      "loss": 0.0723,
      "step": 5177,
      "task_loss": 0.1453644037246704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942734666888551,
      "compression_loss": 0.0,
      "distillation_loss": 0.07094942778348923,
      "epoch": 4.92,
      "learning_rate": 1.0182372451958943e-05,
      "loss": 0.076,
      "step": 5178,
      "task_loss": 0.12134627997875214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942787559742146,
      "compression_loss": 0.0,
      "distillation_loss": 0.11129570007324219,
      "epoch": 4.92,
      "learning_rate": 1.0173791854767155e-05,
      "loss": 0.1065,
      "step": 5179,
      "task_loss": 0.06368857622146606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942840420016218,
      "compression_loss": 0.0,
      "distillation_loss": 0.04373692348599434,
      "epoch": 4.92,
      "learning_rate": 1.0165213950827473e-05,
      "loss": 0.0422,
      "step": 5180,
      "task_loss": 0.028108179569244385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942893247720808,
      "compression_loss": 0.0,
      "distillation_loss": 0.06466871500015259,
      "epoch": 4.92,
      "learning_rate": 1.015663874169813e-05,
      "loss": 0.0863,
      "step": 5181,
      "task_loss": 0.28089067339897156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794294604286595,
      "compression_loss": 0.0,
      "distillation_loss": 0.04316772520542145,
      "epoch": 4.92,
      "learning_rate": 1.0148066228936836e-05,
      "loss": 0.0488,
      "step": 5182,
      "task_loss": 0.09936368465423584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7942998805461682,
      "compression_loss": 0.0,
      "distillation_loss": 0.05181555822491646,
      "epoch": 4.92,
      "learning_rate": 1.013949641410085e-05,
      "loss": 0.0584,
      "step": 5183,
      "task_loss": 0.117317795753479
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794305153551804,
      "compression_loss": 0.0,
      "distillation_loss": 0.020150721073150635,
      "epoch": 4.92,
      "learning_rate": 1.0130929298746909e-05,
      "loss": 0.0317,
      "step": 5184,
      "task_loss": 0.1358027458190918
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943104233045062,
      "compression_loss": 0.0,
      "distillation_loss": 0.06334016472101212,
      "epoch": 4.92,
      "learning_rate": 1.0122364884431259e-05,
      "loss": 0.0633,
      "step": 5185,
      "task_loss": 0.062485143542289734
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943156898052783,
      "compression_loss": 0.0,
      "distillation_loss": 0.05838319659233093,
      "epoch": 4.92,
      "learning_rate": 1.011380317270969e-05,
      "loss": 0.0593,
      "step": 5186,
      "task_loss": 0.06781870126724243
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943209530551243,
      "compression_loss": 0.0,
      "distillation_loss": 0.033147186040878296,
      "epoch": 4.93,
      "learning_rate": 1.0105244165137467e-05,
      "loss": 0.0457,
      "step": 5187,
      "task_loss": 0.15893574059009552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943262130550476,
      "compression_loss": 0.0,
      "distillation_loss": 0.026804693043231964,
      "epoch": 4.93,
      "learning_rate": 1.0096687863269368e-05,
      "loss": 0.0248,
      "step": 5188,
      "task_loss": 0.006701400503516197
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943314698060518,
      "compression_loss": 0.0,
      "distillation_loss": 0.04605764150619507,
      "epoch": 4.93,
      "learning_rate": 1.0088134268659693e-05,
      "loss": 0.0498,
      "step": 5189,
      "task_loss": 0.08354970067739487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794336723309141,
      "compression_loss": 0.0,
      "distillation_loss": 0.033246997743844986,
      "epoch": 4.93,
      "learning_rate": 1.0079583382862254e-05,
      "loss": 0.0326,
      "step": 5190,
      "task_loss": 0.026343755424022675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943419735653184,
      "compression_loss": 0.0,
      "distillation_loss": 0.06967267394065857,
      "epoch": 4.93,
      "learning_rate": 1.0071035207430352e-05,
      "loss": 0.0715,
      "step": 5191,
      "task_loss": 0.08810234814882278
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943472205755882,
      "compression_loss": 0.0,
      "distillation_loss": 0.02802272140979767,
      "epoch": 4.93,
      "learning_rate": 1.0062489743916792e-05,
      "loss": 0.0287,
      "step": 5192,
      "task_loss": 0.03505971282720566
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943524643409535,
      "compression_loss": 0.0,
      "distillation_loss": 0.07618989050388336,
      "epoch": 4.93,
      "learning_rate": 1.005394699387392e-05,
      "loss": 0.0715,
      "step": 5193,
      "task_loss": 0.029166080057621002
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943577048624184,
      "compression_loss": 0.0,
      "distillation_loss": 0.033866822719573975,
      "epoch": 4.93,
      "learning_rate": 1.0045406958853556e-05,
      "loss": 0.0316,
      "step": 5194,
      "task_loss": 0.010730434209108353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943629421409865,
      "compression_loss": 0.0,
      "distillation_loss": 0.03229956328868866,
      "epoch": 4.93,
      "learning_rate": 1.0036869640407031e-05,
      "loss": 0.0305,
      "step": 5195,
      "task_loss": 0.01465969905257225
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943681761776614,
      "compression_loss": 0.0,
      "distillation_loss": 0.040657661855220795,
      "epoch": 4.93,
      "learning_rate": 1.002833504008521e-05,
      "loss": 0.0535,
      "step": 5196,
      "task_loss": 0.16867557168006897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943734069734468,
      "compression_loss": 0.0,
      "distillation_loss": 0.12660034000873566,
      "epoch": 4.94,
      "learning_rate": 1.0019803159438423e-05,
      "loss": 0.1192,
      "step": 5197,
      "task_loss": 0.05233707278966904
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943786345293464,
      "compression_loss": 0.0,
      "distillation_loss": 0.24157142639160156,
      "epoch": 4.94,
      "learning_rate": 1.0011274000016547e-05,
      "loss": 0.237,
      "step": 5198,
      "task_loss": 0.19573716819286346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943838588463639,
      "compression_loss": 0.0,
      "distillation_loss": 0.10286252945661545,
      "epoch": 4.94,
      "learning_rate": 1.0002747563368928e-05,
      "loss": 0.1052,
      "step": 5199,
      "task_loss": 0.12575237452983856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794389079925503,
      "compression_loss": 0.0,
      "distillation_loss": 0.1292041838169098,
      "epoch": 4.94,
      "learning_rate": 9.994223851044449e-06,
      "loss": 0.1243,
      "step": 5200,
      "task_loss": 0.07981815189123154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943942977677673,
      "compression_loss": 0.0,
      "distillation_loss": 0.04530710726976395,
      "epoch": 4.94,
      "learning_rate": 9.985702864591477e-06,
      "loss": 0.0477,
      "step": 5201,
      "task_loss": 0.06911186873912811
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7943995123741605,
      "compression_loss": 0.0,
      "distillation_loss": 0.04243285953998566,
      "epoch": 4.94,
      "learning_rate": 9.977184605557877e-06,
      "loss": 0.0385,
      "step": 5202,
      "task_loss": 0.003344038501381874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944047237456864,
      "compression_loss": 0.0,
      "distillation_loss": 0.01720603182911873,
      "epoch": 4.94,
      "learning_rate": 9.968669075491054e-06,
      "loss": 0.0204,
      "step": 5203,
      "task_loss": 0.04868287593126297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944099318833486,
      "compression_loss": 0.0,
      "distillation_loss": 0.02283567562699318,
      "epoch": 4.94,
      "learning_rate": 9.960156275937887e-06,
      "loss": 0.0279,
      "step": 5204,
      "task_loss": 0.07395943999290466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944151367881507,
      "compression_loss": 0.0,
      "distillation_loss": 0.012866092845797539,
      "epoch": 4.94,
      "learning_rate": 9.951646208444756e-06,
      "loss": 0.0119,
      "step": 5205,
      "task_loss": 0.0030320733785629272
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944203384610965,
      "compression_loss": 0.0,
      "distillation_loss": 0.03689184784889221,
      "epoch": 4.94,
      "learning_rate": 9.943138874557564e-06,
      "loss": 0.0415,
      "step": 5206,
      "task_loss": 0.0826156958937645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944255369031896,
      "compression_loss": 0.0,
      "distillation_loss": 0.01845708116889,
      "epoch": 4.94,
      "learning_rate": 9.934634275821722e-06,
      "loss": 0.0175,
      "step": 5207,
      "task_loss": 0.009108863770961761
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944307321154338,
      "compression_loss": 0.0,
      "distillation_loss": 0.01570175029337406,
      "epoch": 4.95,
      "learning_rate": 9.926132413782116e-06,
      "loss": 0.0147,
      "step": 5208,
      "task_loss": 0.006071802228689194
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944359240988327,
      "compression_loss": 0.0,
      "distillation_loss": 0.013230700977146626,
      "epoch": 4.95,
      "learning_rate": 9.917633289983144e-06,
      "loss": 0.0125,
      "step": 5209,
      "task_loss": 0.0057407524436712265
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79444111285439,
      "compression_loss": 0.0,
      "distillation_loss": 0.035949304699897766,
      "epoch": 4.95,
      "learning_rate": 9.909136905968736e-06,
      "loss": 0.0461,
      "step": 5210,
      "task_loss": 0.13782669603824615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944462983831093,
      "compression_loss": 0.0,
      "distillation_loss": 0.03853258118033409,
      "epoch": 4.95,
      "learning_rate": 9.900643263282283e-06,
      "loss": 0.0386,
      "step": 5211,
      "task_loss": 0.03873224928975105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944514806859944,
      "compression_loss": 0.0,
      "distillation_loss": 0.036589786410331726,
      "epoch": 4.95,
      "learning_rate": 9.892152363466691e-06,
      "loss": 0.036,
      "step": 5212,
      "task_loss": 0.03055756166577339
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944566597640489,
      "compression_loss": 0.0,
      "distillation_loss": 0.03954382613301277,
      "epoch": 4.95,
      "learning_rate": 9.883664208064394e-06,
      "loss": 0.0415,
      "step": 5213,
      "task_loss": 0.05872972309589386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944618356182765,
      "compression_loss": 0.0,
      "distillation_loss": 0.0241532102227211,
      "epoch": 4.95,
      "learning_rate": 9.875178798617286e-06,
      "loss": 0.0284,
      "step": 5214,
      "task_loss": 0.06614520400762558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794467008249681,
      "compression_loss": 0.0,
      "distillation_loss": 0.021967854350805283,
      "epoch": 4.95,
      "learning_rate": 9.866696136666798e-06,
      "loss": 0.0208,
      "step": 5215,
      "task_loss": 0.010009793564677238
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944721776592659,
      "compression_loss": 0.0,
      "distillation_loss": 0.02724001184105873,
      "epoch": 4.95,
      "learning_rate": 9.858216223753833e-06,
      "loss": 0.0286,
      "step": 5216,
      "task_loss": 0.04124392196536064
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794477343848035,
      "compression_loss": 0.0,
      "distillation_loss": 0.027787886559963226,
      "epoch": 4.95,
      "learning_rate": 9.849739061418823e-06,
      "loss": 0.0329,
      "step": 5217,
      "task_loss": 0.07907851040363312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794482506816992,
      "compression_loss": 0.0,
      "distillation_loss": 0.08973988890647888,
      "epoch": 4.96,
      "learning_rate": 9.841264651201676e-06,
      "loss": 0.1021,
      "step": 5218,
      "task_loss": 0.21334464848041534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944876665671404,
      "compression_loss": 0.0,
      "distillation_loss": 0.06084804609417915,
      "epoch": 4.96,
      "learning_rate": 9.832792994641801e-06,
      "loss": 0.0676,
      "step": 5219,
      "task_loss": 0.12829428911209106
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944928230994841,
      "compression_loss": 0.0,
      "distillation_loss": 0.01122710295021534,
      "epoch": 4.96,
      "learning_rate": 9.824324093278134e-06,
      "loss": 0.0247,
      "step": 5220,
      "task_loss": 0.1459704488515854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7944979764150266,
      "compression_loss": 0.0,
      "distillation_loss": 0.06464344263076782,
      "epoch": 4.96,
      "learning_rate": 9.815857948649082e-06,
      "loss": 0.0686,
      "step": 5221,
      "task_loss": 0.10410156100988388
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945031265147717,
      "compression_loss": 0.0,
      "distillation_loss": 0.01737487129867077,
      "epoch": 4.96,
      "learning_rate": 9.807394562292551e-06,
      "loss": 0.0214,
      "step": 5222,
      "task_loss": 0.05750443786382675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945082733997232,
      "compression_loss": 0.0,
      "distillation_loss": 0.04283272475004196,
      "epoch": 4.96,
      "learning_rate": 9.798933935745968e-06,
      "loss": 0.0468,
      "step": 5223,
      "task_loss": 0.08252625167369843
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945134170708845,
      "compression_loss": 0.0,
      "distillation_loss": 0.029033223167061806,
      "epoch": 4.96,
      "learning_rate": 9.790476070546257e-06,
      "loss": 0.0299,
      "step": 5224,
      "task_loss": 0.03801552951335907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945185575292595,
      "compression_loss": 0.0,
      "distillation_loss": 0.03287568688392639,
      "epoch": 4.96,
      "learning_rate": 9.782020968229813e-06,
      "loss": 0.038,
      "step": 5225,
      "task_loss": 0.08447042852640152
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945236947758517,
      "compression_loss": 0.0,
      "distillation_loss": 0.05485154315829277,
      "epoch": 4.96,
      "learning_rate": 9.77356863033254e-06,
      "loss": 0.0504,
      "step": 5226,
      "task_loss": 0.010427333414554596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794528828811665,
      "compression_loss": 0.0,
      "distillation_loss": 0.030403347685933113,
      "epoch": 4.96,
      "learning_rate": 9.765119058389872e-06,
      "loss": 0.0276,
      "step": 5227,
      "task_loss": 0.0026769302785396576
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945339596377028,
      "compression_loss": 0.0,
      "distillation_loss": 0.023587405681610107,
      "epoch": 4.96,
      "learning_rate": 9.756672253936694e-06,
      "loss": 0.0219,
      "step": 5228,
      "task_loss": 0.007051067426800728
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945390872549691,
      "compression_loss": 0.0,
      "distillation_loss": 0.13783934712409973,
      "epoch": 4.97,
      "learning_rate": 9.748228218507408e-06,
      "loss": 0.1264,
      "step": 5229,
      "task_loss": 0.02342773601412773
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945442116644673,
      "compression_loss": 0.0,
      "distillation_loss": 0.02738800458610058,
      "epoch": 4.97,
      "learning_rate": 9.739786953635924e-06,
      "loss": 0.0255,
      "step": 5230,
      "task_loss": 0.008501600474119186
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945493328672013,
      "compression_loss": 0.0,
      "distillation_loss": 0.1387009620666504,
      "epoch": 4.97,
      "learning_rate": 9.731348460855638e-06,
      "loss": 0.1468,
      "step": 5231,
      "task_loss": 0.21995283663272858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945544508641748,
      "compression_loss": 0.0,
      "distillation_loss": 0.02429133839905262,
      "epoch": 4.97,
      "learning_rate": 9.722912741699427e-06,
      "loss": 0.0272,
      "step": 5232,
      "task_loss": 0.05331313982605934
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945595656563912,
      "compression_loss": 0.0,
      "distillation_loss": 0.046510256826877594,
      "epoch": 4.97,
      "learning_rate": 9.714479797699694e-06,
      "loss": 0.0428,
      "step": 5233,
      "task_loss": 0.009763523936271667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945646772448545,
      "compression_loss": 0.0,
      "distillation_loss": 0.025538455694913864,
      "epoch": 4.97,
      "learning_rate": 9.706049630388329e-06,
      "loss": 0.0352,
      "step": 5234,
      "task_loss": 0.12184809148311615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945697856305681,
      "compression_loss": 0.0,
      "distillation_loss": 0.11467162519693375,
      "epoch": 4.97,
      "learning_rate": 9.697622241296705e-06,
      "loss": 0.1134,
      "step": 5235,
      "task_loss": 0.10236775875091553
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945748908145359,
      "compression_loss": 0.0,
      "distillation_loss": 0.0173861812800169,
      "epoch": 4.97,
      "learning_rate": 9.68919763195569e-06,
      "loss": 0.0161,
      "step": 5236,
      "task_loss": 0.004365898668766022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945799927977615,
      "compression_loss": 0.0,
      "distillation_loss": 0.014998473227024078,
      "epoch": 4.97,
      "learning_rate": 9.680775803895672e-06,
      "loss": 0.0141,
      "step": 5237,
      "task_loss": 0.005868380889296532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945850915812486,
      "compression_loss": 0.0,
      "distillation_loss": 0.04273030161857605,
      "epoch": 4.97,
      "learning_rate": 9.672356758646506e-06,
      "loss": 0.0546,
      "step": 5238,
      "task_loss": 0.1609753668308258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945901871660008,
      "compression_loss": 0.0,
      "distillation_loss": 0.03701045364141464,
      "epoch": 4.98,
      "learning_rate": 9.663940497737556e-06,
      "loss": 0.0345,
      "step": 5239,
      "task_loss": 0.01145661249756813
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7945952795530219,
      "compression_loss": 0.0,
      "distillation_loss": 0.01707928627729416,
      "epoch": 4.98,
      "learning_rate": 9.655527022697652e-06,
      "loss": 0.0161,
      "step": 5240,
      "task_loss": 0.00770198181271553
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946003687433155,
      "compression_loss": 0.0,
      "distillation_loss": 0.034336067736148834,
      "epoch": 4.98,
      "learning_rate": 9.647116335055187e-06,
      "loss": 0.0393,
      "step": 5241,
      "task_loss": 0.08418752998113632
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946054547378854,
      "compression_loss": 0.0,
      "distillation_loss": 0.026298392564058304,
      "epoch": 4.98,
      "learning_rate": 9.638708436337976e-06,
      "loss": 0.0248,
      "step": 5242,
      "task_loss": 0.01096322201192379
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946105375377351,
      "compression_loss": 0.0,
      "distillation_loss": 0.03858939930796623,
      "epoch": 4.98,
      "learning_rate": 9.630303328073352e-06,
      "loss": 0.0371,
      "step": 5243,
      "task_loss": 0.02343849278986454
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946156171438685,
      "compression_loss": 0.0,
      "distillation_loss": 0.029876116663217545,
      "epoch": 4.98,
      "learning_rate": 9.621901011788157e-06,
      "loss": 0.0363,
      "step": 5244,
      "task_loss": 0.09448274224996567
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794620693557289,
      "compression_loss": 0.0,
      "distillation_loss": 0.07018247991800308,
      "epoch": 4.98,
      "learning_rate": 9.613501489008705e-06,
      "loss": 0.08,
      "step": 5245,
      "task_loss": 0.1685934066772461
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946257667790007,
      "compression_loss": 0.0,
      "distillation_loss": 0.08023358136415482,
      "epoch": 4.98,
      "learning_rate": 9.605104761260809e-06,
      "loss": 0.0914,
      "step": 5246,
      "task_loss": 0.19211246073246002
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946308368100068,
      "compression_loss": 0.0,
      "distillation_loss": 0.020654667168855667,
      "epoch": 4.98,
      "learning_rate": 9.596710830069769e-06,
      "loss": 0.0201,
      "step": 5247,
      "task_loss": 0.014910630881786346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946359036513112,
      "compression_loss": 0.0,
      "distillation_loss": 0.07811079919338226,
      "epoch": 4.98,
      "learning_rate": 9.588319696960396e-06,
      "loss": 0.0898,
      "step": 5248,
      "task_loss": 0.19523471593856812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946409673039178,
      "compression_loss": 0.0,
      "distillation_loss": 0.14166250824928284,
      "epoch": 4.98,
      "learning_rate": 9.579931363456968e-06,
      "loss": 0.1623,
      "step": 5249,
      "task_loss": 0.3483670949935913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946460277688299,
      "compression_loss": 0.0,
      "distillation_loss": 0.11959753930568695,
      "epoch": 4.99,
      "learning_rate": 9.571545831083267e-06,
      "loss": 0.1251,
      "step": 5250,
      "task_loss": 0.17435845732688904
    },
    {
      "epoch": 4.99,
      "eval_accuracy": 0.9013761467889908,
      "eval_loss": 0.40377190709114075,
      "eval_runtime": 18.2101,
      "eval_samples_per_second": 47.886,
      "eval_steps_per_second": 5.986,
      "step": 5250
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946510850470513,
      "compression_loss": 0.0,
      "distillation_loss": 0.09491365402936935,
      "epoch": 4.99,
      "learning_rate": 9.563163101362582e-06,
      "loss": 0.0985,
      "step": 5251,
      "task_loss": 0.13050074875354767
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946561391395859,
      "compression_loss": 0.0,
      "distillation_loss": 0.043609265238046646,
      "epoch": 4.99,
      "learning_rate": 9.554783175817667e-06,
      "loss": 0.0419,
      "step": 5252,
      "task_loss": 0.0262824147939682
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946611900474371,
      "compression_loss": 0.0,
      "distillation_loss": 0.03674563765525818,
      "epoch": 4.99,
      "learning_rate": 9.546406055970769e-06,
      "loss": 0.0376,
      "step": 5253,
      "task_loss": 0.045376792550086975
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946662377716087,
      "compression_loss": 0.0,
      "distillation_loss": 0.015710389241576195,
      "epoch": 4.99,
      "learning_rate": 9.538031743343628e-06,
      "loss": 0.0144,
      "step": 5254,
      "task_loss": 0.0030107852071523666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946712823131045,
      "compression_loss": 0.0,
      "distillation_loss": 0.018925126641988754,
      "epoch": 4.99,
      "learning_rate": 9.529660239457497e-06,
      "loss": 0.0175,
      "step": 5255,
      "task_loss": 0.005088077858090401
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794676323672928,
      "compression_loss": 0.0,
      "distillation_loss": 0.08420512825250626,
      "epoch": 4.99,
      "learning_rate": 9.521291545833086e-06,
      "loss": 0.0801,
      "step": 5256,
      "task_loss": 0.0430089607834816
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946813618520829,
      "compression_loss": 0.0,
      "distillation_loss": 0.019614659249782562,
      "epoch": 4.99,
      "learning_rate": 9.512925663990608e-06,
      "loss": 0.0183,
      "step": 5257,
      "task_loss": 0.006246289238333702
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946863968515729,
      "compression_loss": 0.0,
      "distillation_loss": 0.029730752110481262,
      "epoch": 4.99,
      "learning_rate": 9.504562595449764e-06,
      "loss": 0.0362,
      "step": 5258,
      "task_loss": 0.0940956100821495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946914286724018,
      "compression_loss": 0.0,
      "distillation_loss": 0.1385657787322998,
      "epoch": 4.99,
      "learning_rate": 9.496202341729765e-06,
      "loss": 0.1312,
      "step": 5259,
      "task_loss": 0.06457454711198807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7946964573155733,
      "compression_loss": 0.0,
      "distillation_loss": 0.02025190182030201,
      "epoch": 5.0,
      "learning_rate": 9.487844904349274e-06,
      "loss": 0.0327,
      "step": 5260,
      "task_loss": 0.1445349156856537
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947014827820907,
      "compression_loss": 0.0,
      "distillation_loss": 0.119037926197052,
      "epoch": 5.0,
      "learning_rate": 9.479490284826459e-06,
      "loss": 0.1162,
      "step": 5261,
      "task_loss": 0.09018512070178986
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947065050729581,
      "compression_loss": 0.0,
      "distillation_loss": 0.054691120982170105,
      "epoch": 5.0,
      "learning_rate": 9.471138484678987e-06,
      "loss": 0.0638,
      "step": 5262,
      "task_loss": 0.14550621807575226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947115241891791,
      "compression_loss": 0.0,
      "distillation_loss": 0.1421840786933899,
      "epoch": 5.0,
      "learning_rate": 9.462789505423997e-06,
      "loss": 0.1377,
      "step": 5263,
      "task_loss": 0.09778082370758057
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947165401317573,
      "compression_loss": 0.0,
      "distillation_loss": 0.039873864501714706,
      "epoch": 5.0,
      "learning_rate": 9.454443348578115e-06,
      "loss": 0.0551,
      "step": 5264,
      "task_loss": 0.19219006597995758
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6151754920258597,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7874999929845908,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947215529016963,
      "compression_loss": 0.0,
      "distillation_loss": 0.021231336519122124,
      "epoch": 5.0,
      "learning_rate": 9.446100015657474e-06,
      "loss": 0.0198,
      "step": 5265,
      "task_loss": 0.007257444318383932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947265625000001,
      "compression_loss": 0.0,
      "distillation_loss": 0.08732712268829346,
      "epoch": 5.0,
      "learning_rate": 9.437759508177665e-06,
      "loss": 0.0898,
      "step": 5266,
      "task_loss": 0.11206232756376266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947315689276719,
      "compression_loss": 0.0,
      "distillation_loss": 0.2404271364212036,
      "epoch": 5.0,
      "learning_rate": 9.429421827653798e-06,
      "loss": 0.2203,
      "step": 5267,
      "task_loss": 0.03922963887453079
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947365721857158,
      "compression_loss": 0.0,
      "distillation_loss": 0.48191341757774353,
      "epoch": 5.0,
      "learning_rate": 9.421086975600438e-06,
      "loss": 0.484,
      "step": 5268,
      "task_loss": 0.5028781294822693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947415722751353,
      "compression_loss": 0.0,
      "distillation_loss": 0.2334894835948944,
      "epoch": 5.0,
      "learning_rate": 9.412754953531663e-06,
      "loss": 0.2161,
      "step": 5269,
      "task_loss": 0.059303127229213715
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947465691969342,
      "compression_loss": 0.0,
      "distillation_loss": 0.17375622689723969,
      "epoch": 5.0,
      "learning_rate": 9.404425762961022e-06,
      "loss": 0.1641,
      "step": 5270,
      "task_loss": 0.07717826217412949
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794751562952116,
      "compression_loss": 0.0,
      "distillation_loss": 0.40218544006347656,
      "epoch": 5.01,
      "learning_rate": 9.396099405401543e-06,
      "loss": 0.3844,
      "step": 5271,
      "task_loss": 0.22383946180343628
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947565535416845,
      "compression_loss": 0.0,
      "distillation_loss": 0.28495174646377563,
      "epoch": 5.01,
      "learning_rate": 9.387775882365763e-06,
      "loss": 0.2911,
      "step": 5272,
      "task_loss": 0.34688353538513184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947615409666433,
      "compression_loss": 0.0,
      "distillation_loss": 0.2111060917377472,
      "epoch": 5.01,
      "learning_rate": 9.379455195365684e-06,
      "loss": 0.2001,
      "step": 5273,
      "task_loss": 0.10126888751983643
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947665252279963,
      "compression_loss": 0.0,
      "distillation_loss": 0.15187914669513702,
      "epoch": 5.01,
      "learning_rate": 9.371137345912786e-06,
      "loss": 0.1592,
      "step": 5274,
      "task_loss": 0.225071519613266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947715063267469,
      "compression_loss": 0.0,
      "distillation_loss": 0.22621603310108185,
      "epoch": 5.01,
      "learning_rate": 9.362822335518063e-06,
      "loss": 0.2191,
      "step": 5275,
      "task_loss": 0.15458367764949799
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947764842638989,
      "compression_loss": 0.0,
      "distillation_loss": 0.11143049597740173,
      "epoch": 5.01,
      "learning_rate": 9.35451016569198e-06,
      "loss": 0.1072,
      "step": 5276,
      "task_loss": 0.06862667948007584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794781459040456,
      "compression_loss": 0.0,
      "distillation_loss": 0.070254385471344,
      "epoch": 5.01,
      "learning_rate": 9.346200837944474e-06,
      "loss": 0.0712,
      "step": 5277,
      "task_loss": 0.07974936068058014
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947864306574219,
      "compression_loss": 0.0,
      "distillation_loss": 0.04427797347307205,
      "epoch": 5.01,
      "learning_rate": 9.337894353784968e-06,
      "loss": 0.04,
      "step": 5278,
      "task_loss": 0.0014720112085342407
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947913991158003,
      "compression_loss": 0.0,
      "distillation_loss": 0.23665665090084076,
      "epoch": 5.01,
      "learning_rate": 9.329590714722388e-06,
      "loss": 0.2457,
      "step": 5279,
      "task_loss": 0.32665759325027466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7947963644165947,
      "compression_loss": 0.0,
      "distillation_loss": 0.18532395362854004,
      "epoch": 5.01,
      "learning_rate": 9.321289922265124e-06,
      "loss": 0.1669,
      "step": 5280,
      "task_loss": 0.0015691444277763367
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794801326560809,
      "compression_loss": 0.0,
      "distillation_loss": 0.20857329666614532,
      "epoch": 5.02,
      "learning_rate": 9.312991977921051e-06,
      "loss": 0.198,
      "step": 5281,
      "task_loss": 0.10306287556886673
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948062855494469,
      "compression_loss": 0.0,
      "distillation_loss": 0.11233891546726227,
      "epoch": 5.02,
      "learning_rate": 9.304696883197542e-06,
      "loss": 0.1079,
      "step": 5282,
      "task_loss": 0.06844587624073029
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948112413835119,
      "compression_loss": 0.0,
      "distillation_loss": 0.15439686179161072,
      "epoch": 5.02,
      "learning_rate": 9.29640463960143e-06,
      "loss": 0.1511,
      "step": 5283,
      "task_loss": 0.12120820581912994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948161940640077,
      "compression_loss": 0.0,
      "distillation_loss": 0.07337729632854462,
      "epoch": 5.02,
      "learning_rate": 9.28811524863904e-06,
      "loss": 0.0865,
      "step": 5284,
      "task_loss": 0.2042730152606964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948211435919381,
      "compression_loss": 0.0,
      "distillation_loss": 0.041931264102458954,
      "epoch": 5.02,
      "learning_rate": 9.279828711816182e-06,
      "loss": 0.0454,
      "step": 5285,
      "task_loss": 0.0765017569065094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948260899683067,
      "compression_loss": 0.0,
      "distillation_loss": 0.21245259046554565,
      "epoch": 5.02,
      "learning_rate": 9.271545030638156e-06,
      "loss": 0.2016,
      "step": 5286,
      "task_loss": 0.10362794995307922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948310331941172,
      "compression_loss": 0.0,
      "distillation_loss": 0.06203429400920868,
      "epoch": 5.02,
      "learning_rate": 9.263264206609726e-06,
      "loss": 0.0577,
      "step": 5287,
      "task_loss": 0.018689151853322983
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948359732703734,
      "compression_loss": 0.0,
      "distillation_loss": 0.07664155215024948,
      "epoch": 5.02,
      "learning_rate": 9.254986241235131e-06,
      "loss": 0.0782,
      "step": 5288,
      "task_loss": 0.09256763756275177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948409101980788,
      "compression_loss": 0.0,
      "distillation_loss": 0.1382894665002823,
      "epoch": 5.02,
      "learning_rate": 9.246711136018124e-06,
      "loss": 0.1313,
      "step": 5289,
      "task_loss": 0.06794591248035431
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948458439782371,
      "compression_loss": 0.0,
      "distillation_loss": 0.18572860956192017,
      "epoch": 5.02,
      "learning_rate": 9.23843889246191e-06,
      "loss": 0.1894,
      "step": 5290,
      "task_loss": 0.22227227687835693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794850774611852,
      "compression_loss": 0.0,
      "distillation_loss": 0.11491254717111588,
      "epoch": 5.02,
      "learning_rate": 9.230169512069168e-06,
      "loss": 0.1096,
      "step": 5291,
      "task_loss": 0.061589665710926056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948557020999274,
      "compression_loss": 0.0,
      "distillation_loss": 0.03495645150542259,
      "epoch": 5.03,
      "learning_rate": 9.221902996342084e-06,
      "loss": 0.0318,
      "step": 5292,
      "task_loss": 0.0034442245960235596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948606264434667,
      "compression_loss": 0.0,
      "distillation_loss": 0.32723939418792725,
      "epoch": 5.03,
      "learning_rate": 9.213639346782315e-06,
      "loss": 0.3206,
      "step": 5293,
      "task_loss": 0.261165052652359
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948655476434737,
      "compression_loss": 0.0,
      "distillation_loss": 0.03939376026391983,
      "epoch": 5.03,
      "learning_rate": 9.20537856489099e-06,
      "loss": 0.0538,
      "step": 5294,
      "task_loss": 0.18324826657772064
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948704657009521,
      "compression_loss": 0.0,
      "distillation_loss": 0.14817285537719727,
      "epoch": 5.03,
      "learning_rate": 9.197120652168706e-06,
      "loss": 0.1403,
      "step": 5295,
      "task_loss": 0.06911684572696686
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948753806169055,
      "compression_loss": 0.0,
      "distillation_loss": 0.0447382926940918,
      "epoch": 5.03,
      "learning_rate": 9.18886561011557e-06,
      "loss": 0.0421,
      "step": 5296,
      "task_loss": 0.01793717034161091
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948802923923376,
      "compression_loss": 0.0,
      "distillation_loss": 0.05252248793840408,
      "epoch": 5.03,
      "learning_rate": 9.180613440231145e-06,
      "loss": 0.0492,
      "step": 5297,
      "task_loss": 0.019047562032938004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948852010282522,
      "compression_loss": 0.0,
      "distillation_loss": 0.10433128476142883,
      "epoch": 5.03,
      "learning_rate": 9.172364144014466e-06,
      "loss": 0.1018,
      "step": 5298,
      "task_loss": 0.07895547151565552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948901065256528,
      "compression_loss": 0.0,
      "distillation_loss": 0.04161163419485092,
      "epoch": 5.03,
      "learning_rate": 9.164117722964077e-06,
      "loss": 0.0458,
      "step": 5299,
      "task_loss": 0.0835949033498764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948950088855433,
      "compression_loss": 0.0,
      "distillation_loss": 0.1584656685590744,
      "epoch": 5.03,
      "learning_rate": 9.155874178577973e-06,
      "loss": 0.1606,
      "step": 5300,
      "task_loss": 0.18005318939685822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7948999081089272,
      "compression_loss": 0.0,
      "distillation_loss": 0.16821911931037903,
      "epoch": 5.03,
      "learning_rate": 9.147633512353618e-06,
      "loss": 0.1554,
      "step": 5301,
      "task_loss": 0.04032587260007858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949048041968082,
      "compression_loss": 0.0,
      "distillation_loss": 0.046078503131866455,
      "epoch": 5.04,
      "learning_rate": 9.139395725787981e-06,
      "loss": 0.0506,
      "step": 5302,
      "task_loss": 0.09137886017560959
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79490969715019,
      "compression_loss": 0.0,
      "distillation_loss": 0.1656574308872223,
      "epoch": 5.04,
      "learning_rate": 9.131160820377505e-06,
      "loss": 0.1672,
      "step": 5303,
      "task_loss": 0.18076612055301666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949145869700764,
      "compression_loss": 0.0,
      "distillation_loss": 0.159349262714386,
      "epoch": 5.04,
      "learning_rate": 9.12292879761809e-06,
      "loss": 0.1652,
      "step": 5304,
      "task_loss": 0.21749325096607208
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.794919473657471,
      "compression_loss": 0.0,
      "distillation_loss": 0.09572987258434296,
      "epoch": 5.04,
      "learning_rate": 9.11469965900512e-06,
      "loss": 0.1055,
      "step": 5305,
      "task_loss": 0.19356384873390198
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949243572133774,
      "compression_loss": 0.0,
      "distillation_loss": 0.06085019186139107,
      "epoch": 5.04,
      "learning_rate": 9.106473406033464e-06,
      "loss": 0.0755,
      "step": 5306,
      "task_loss": 0.2072937786579132
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949292376387993,
      "compression_loss": 0.0,
      "distillation_loss": 0.09842538833618164,
      "epoch": 5.04,
      "learning_rate": 9.098250040197462e-06,
      "loss": 0.113,
      "step": 5307,
      "task_loss": 0.24399572610855103
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949341149347405,
      "compression_loss": 0.0,
      "distillation_loss": 0.1453973799943924,
      "epoch": 5.04,
      "learning_rate": 9.090029562990911e-06,
      "loss": 0.1474,
      "step": 5308,
      "task_loss": 0.16563892364501953
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949389891022046,
      "compression_loss": 0.0,
      "distillation_loss": 0.04174884408712387,
      "epoch": 5.04,
      "learning_rate": 9.081811975907126e-06,
      "loss": 0.0491,
      "step": 5309,
      "task_loss": 0.11511567234992981
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949438601421954,
      "compression_loss": 0.0,
      "distillation_loss": 0.02893233485519886,
      "epoch": 5.04,
      "learning_rate": 9.07359728043885e-06,
      "loss": 0.0264,
      "step": 5310,
      "task_loss": 0.003949105739593506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949487280557164,
      "compression_loss": 0.0,
      "distillation_loss": 0.048204317688941956,
      "epoch": 5.04,
      "learning_rate": 9.065385478078337e-06,
      "loss": 0.0441,
      "step": 5311,
      "task_loss": 0.007329225540161133
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949535928437713,
      "compression_loss": 0.0,
      "distillation_loss": 0.043466269969940186,
      "epoch": 5.04,
      "learning_rate": 9.057176570317285e-06,
      "loss": 0.0406,
      "step": 5312,
      "task_loss": 0.015056092292070389
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949584545073639,
      "compression_loss": 0.0,
      "distillation_loss": 0.21660102903842926,
      "epoch": 5.05,
      "learning_rate": 9.0489705586469e-06,
      "loss": 0.2129,
      "step": 5313,
      "task_loss": 0.17953038215637207
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949633130474979,
      "compression_loss": 0.0,
      "distillation_loss": 0.11881037801504135,
      "epoch": 5.05,
      "learning_rate": 9.040767444557836e-06,
      "loss": 0.1213,
      "step": 5314,
      "task_loss": 0.14401260018348694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949681684651768,
      "compression_loss": 0.0,
      "distillation_loss": 0.11176039278507233,
      "epoch": 5.05,
      "learning_rate": 9.032567229540212e-06,
      "loss": 0.1134,
      "step": 5315,
      "task_loss": 0.12841668725013733
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949730207614044,
      "compression_loss": 0.0,
      "distillation_loss": 0.14567351341247559,
      "epoch": 5.05,
      "learning_rate": 9.024369915083664e-06,
      "loss": 0.1399,
      "step": 5316,
      "task_loss": 0.0879807248711586
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949778699371844,
      "compression_loss": 0.0,
      "distillation_loss": 0.05770561844110489,
      "epoch": 5.05,
      "learning_rate": 9.01617550267726e-06,
      "loss": 0.0608,
      "step": 5317,
      "task_loss": 0.08822789788246155
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949827159935204,
      "compression_loss": 0.0,
      "distillation_loss": 0.10660059005022049,
      "epoch": 5.05,
      "learning_rate": 9.007983993809543e-06,
      "loss": 0.0995,
      "step": 5318,
      "task_loss": 0.03520110249519348
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949875589314163,
      "compression_loss": 0.0,
      "distillation_loss": 0.06890207529067993,
      "epoch": 5.05,
      "learning_rate": 8.999795389968554e-06,
      "loss": 0.0647,
      "step": 5319,
      "task_loss": 0.027203164994716644
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949923987518754,
      "compression_loss": 0.0,
      "distillation_loss": 0.045612677931785583,
      "epoch": 5.05,
      "learning_rate": 8.9916096926418e-06,
      "loss": 0.048,
      "step": 5320,
      "task_loss": 0.06988806277513504
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7949972354559018,
      "compression_loss": 0.0,
      "distillation_loss": 0.054146356880664825,
      "epoch": 5.05,
      "learning_rate": 8.983426903316242e-06,
      "loss": 0.0589,
      "step": 5321,
      "task_loss": 0.10151663422584534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950020690444989,
      "compression_loss": 0.0,
      "distillation_loss": 0.09348467737436295,
      "epoch": 5.05,
      "learning_rate": 8.975247023478314e-06,
      "loss": 0.0887,
      "step": 5322,
      "task_loss": 0.045579515397548676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950068995186704,
      "compression_loss": 0.0,
      "distillation_loss": 0.04567846655845642,
      "epoch": 5.06,
      "learning_rate": 8.967070054613949e-06,
      "loss": 0.048,
      "step": 5323,
      "task_loss": 0.06916110217571259
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950117268794201,
      "compression_loss": 0.0,
      "distillation_loss": 0.03903430327773094,
      "epoch": 5.06,
      "learning_rate": 8.958895998208524e-06,
      "loss": 0.0357,
      "step": 5324,
      "task_loss": 0.005971364676952362
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950165511277517,
      "compression_loss": 0.0,
      "distillation_loss": 0.16417047381401062,
      "epoch": 5.06,
      "learning_rate": 8.95072485574689e-06,
      "loss": 0.165,
      "step": 5325,
      "task_loss": 0.17254981398582458
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950213722646687,
      "compression_loss": 0.0,
      "distillation_loss": 0.05708067864179611,
      "epoch": 5.06,
      "learning_rate": 8.942556628713386e-06,
      "loss": 0.0848,
      "step": 5326,
      "task_loss": 0.33394309878349304
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795026190291175,
      "compression_loss": 0.0,
      "distillation_loss": 0.10749347507953644,
      "epoch": 5.06,
      "learning_rate": 8.934391318591794e-06,
      "loss": 0.1102,
      "step": 5327,
      "task_loss": 0.13443301618099213
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950310052082741,
      "compression_loss": 0.0,
      "distillation_loss": 0.07780588418245316,
      "epoch": 5.06,
      "learning_rate": 8.926228926865404e-06,
      "loss": 0.0898,
      "step": 5328,
      "task_loss": 0.19749955832958221
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950358170169698,
      "compression_loss": 0.0,
      "distillation_loss": 0.1407744139432907,
      "epoch": 5.06,
      "learning_rate": 8.918069455016931e-06,
      "loss": 0.1353,
      "step": 5329,
      "task_loss": 0.08600475639104843
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950406257182657,
      "compression_loss": 0.0,
      "distillation_loss": 0.09022311121225357,
      "epoch": 5.06,
      "learning_rate": 8.9099129045286e-06,
      "loss": 0.0966,
      "step": 5330,
      "task_loss": 0.15400820970535278
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950454313131655,
      "compression_loss": 0.0,
      "distillation_loss": 0.18497398495674133,
      "epoch": 5.06,
      "learning_rate": 8.901759276882082e-06,
      "loss": 0.2032,
      "step": 5331,
      "task_loss": 0.36690419912338257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795050233802673,
      "compression_loss": 0.0,
      "distillation_loss": 0.04657813161611557,
      "epoch": 5.06,
      "learning_rate": 8.893608573558515e-06,
      "loss": 0.0449,
      "step": 5332,
      "task_loss": 0.029313866049051285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950550331877917,
      "compression_loss": 0.0,
      "distillation_loss": 0.22918027639389038,
      "epoch": 5.06,
      "learning_rate": 8.885460796038525e-06,
      "loss": 0.2344,
      "step": 5333,
      "task_loss": 0.2814333438873291
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950598294695255,
      "compression_loss": 0.0,
      "distillation_loss": 0.02877352014183998,
      "epoch": 5.07,
      "learning_rate": 8.877315945802191e-06,
      "loss": 0.0269,
      "step": 5334,
      "task_loss": 0.010271269828081131
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950646226488778,
      "compression_loss": 0.0,
      "distillation_loss": 0.039790619164705276,
      "epoch": 5.07,
      "learning_rate": 8.869174024329063e-06,
      "loss": 0.0367,
      "step": 5335,
      "task_loss": 0.00851232185959816
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950694127268525,
      "compression_loss": 0.0,
      "distillation_loss": 0.05201128497719765,
      "epoch": 5.07,
      "learning_rate": 8.861035033098144e-06,
      "loss": 0.0603,
      "step": 5336,
      "task_loss": 0.13527747988700867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950741997044531,
      "compression_loss": 0.0,
      "distillation_loss": 0.02241450734436512,
      "epoch": 5.07,
      "learning_rate": 8.852898973587953e-06,
      "loss": 0.0208,
      "step": 5337,
      "task_loss": 0.005801070481538773
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950789835826836,
      "compression_loss": 0.0,
      "distillation_loss": 0.12234519422054291,
      "epoch": 5.07,
      "learning_rate": 8.844765847276432e-06,
      "loss": 0.1178,
      "step": 5338,
      "task_loss": 0.07687139511108398
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950837643625474,
      "compression_loss": 0.0,
      "distillation_loss": 0.042783450335264206,
      "epoch": 5.07,
      "learning_rate": 8.836635655640988e-06,
      "loss": 0.0391,
      "step": 5339,
      "task_loss": 0.0057541318237781525
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950885420450482,
      "compression_loss": 0.0,
      "distillation_loss": 0.20203274488449097,
      "epoch": 5.07,
      "learning_rate": 8.828508400158531e-06,
      "loss": 0.1948,
      "step": 5340,
      "task_loss": 0.13001954555511475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950933166311898,
      "compression_loss": 0.0,
      "distillation_loss": 0.03790685907006264,
      "epoch": 5.07,
      "learning_rate": 8.820384082305403e-06,
      "loss": 0.0417,
      "step": 5341,
      "task_loss": 0.07593842595815659
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7950980881219757,
      "compression_loss": 0.0,
      "distillation_loss": 0.02947581559419632,
      "epoch": 5.07,
      "learning_rate": 8.812262703557431e-06,
      "loss": 0.027,
      "step": 5342,
      "task_loss": 0.004847317934036255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951028565184098,
      "compression_loss": 0.0,
      "distillation_loss": 0.03460695222020149,
      "epoch": 5.07,
      "learning_rate": 8.804144265389888e-06,
      "loss": 0.0316,
      "step": 5343,
      "task_loss": 0.004228517413139343
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951076218214957,
      "compression_loss": 0.0,
      "distillation_loss": 0.1003609448671341,
      "epoch": 5.08,
      "learning_rate": 8.796028769277542e-06,
      "loss": 0.0966,
      "step": 5344,
      "task_loss": 0.06230350583791733
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951123840322372,
      "compression_loss": 0.0,
      "distillation_loss": 0.06244899705052376,
      "epoch": 5.08,
      "learning_rate": 8.78791621669462e-06,
      "loss": 0.0731,
      "step": 5345,
      "task_loss": 0.16876111924648285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951171431516376,
      "compression_loss": 0.0,
      "distillation_loss": 0.030795224010944366,
      "epoch": 5.08,
      "learning_rate": 8.77980660911479e-06,
      "loss": 0.0394,
      "step": 5346,
      "task_loss": 0.11702409386634827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795121899180701,
      "compression_loss": 0.0,
      "distillation_loss": 0.12711569666862488,
      "epoch": 5.08,
      "learning_rate": 8.771699948011202e-06,
      "loss": 0.1325,
      "step": 5347,
      "task_loss": 0.18047887086868286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951266521204309,
      "compression_loss": 0.0,
      "distillation_loss": 0.05044536665081978,
      "epoch": 5.08,
      "learning_rate": 8.763596234856482e-06,
      "loss": 0.0528,
      "step": 5348,
      "task_loss": 0.07405995577573776
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795131401971831,
      "compression_loss": 0.0,
      "distillation_loss": 0.08389607816934586,
      "epoch": 5.08,
      "learning_rate": 8.7554954711227e-06,
      "loss": 0.0782,
      "step": 5349,
      "task_loss": 0.02667861431837082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951361487359049,
      "compression_loss": 0.0,
      "distillation_loss": 0.06966320425271988,
      "epoch": 5.08,
      "learning_rate": 8.747397658281395e-06,
      "loss": 0.0719,
      "step": 5350,
      "task_loss": 0.09200514107942581
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951408924136564,
      "compression_loss": 0.0,
      "distillation_loss": 0.04684102535247803,
      "epoch": 5.08,
      "learning_rate": 8.739302797803585e-06,
      "loss": 0.0432,
      "step": 5351,
      "task_loss": 0.010602116584777832
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951456330060892,
      "compression_loss": 0.0,
      "distillation_loss": 0.024965034797787666,
      "epoch": 5.08,
      "learning_rate": 8.731210891159736e-06,
      "loss": 0.0232,
      "step": 5352,
      "task_loss": 0.0075826942920684814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951503705142069,
      "compression_loss": 0.0,
      "distillation_loss": 0.04389042407274246,
      "epoch": 5.08,
      "learning_rate": 8.72312193981977e-06,
      "loss": 0.0407,
      "step": 5353,
      "task_loss": 0.011852305382490158
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951551049390132,
      "compression_loss": 0.0,
      "distillation_loss": 0.1792694330215454,
      "epoch": 5.08,
      "learning_rate": 8.715035945253095e-06,
      "loss": 0.1675,
      "step": 5354,
      "task_loss": 0.06159057095646858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951598362815118,
      "compression_loss": 0.0,
      "distillation_loss": 0.07054481655359268,
      "epoch": 5.09,
      "learning_rate": 8.706952908928576e-06,
      "loss": 0.0744,
      "step": 5355,
      "task_loss": 0.1091315969824791
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951645645427063,
      "compression_loss": 0.0,
      "distillation_loss": 0.07862917333841324,
      "epoch": 5.09,
      "learning_rate": 8.69887283231453e-06,
      "loss": 0.0713,
      "step": 5356,
      "task_loss": 0.005761042237281799
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951692897236006,
      "compression_loss": 0.0,
      "distillation_loss": 0.07857898622751236,
      "epoch": 5.09,
      "learning_rate": 8.690795716878733e-06,
      "loss": 0.0842,
      "step": 5357,
      "task_loss": 0.1351117193698883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951740118251982,
      "compression_loss": 0.0,
      "distillation_loss": 0.025391805917024612,
      "epoch": 5.09,
      "learning_rate": 8.68272156408845e-06,
      "loss": 0.0238,
      "step": 5358,
      "task_loss": 0.00902603566646576
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951787308485028,
      "compression_loss": 0.0,
      "distillation_loss": 0.033960044384002686,
      "epoch": 5.09,
      "learning_rate": 8.67465037541038e-06,
      "loss": 0.0314,
      "step": 5359,
      "task_loss": 0.008060634136199951
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795183446794518,
      "compression_loss": 0.0,
      "distillation_loss": 0.015219409950077534,
      "epoch": 5.09,
      "learning_rate": 8.666582152310681e-06,
      "loss": 0.0192,
      "step": 5360,
      "task_loss": 0.05521482601761818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951881596642477,
      "compression_loss": 0.0,
      "distillation_loss": 0.04451043903827667,
      "epoch": 5.09,
      "learning_rate": 8.658516896255008e-06,
      "loss": 0.0419,
      "step": 5361,
      "task_loss": 0.01849602535367012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7951928694586955,
      "compression_loss": 0.0,
      "distillation_loss": 0.10168831050395966,
      "epoch": 5.09,
      "learning_rate": 8.650454608708434e-06,
      "loss": 0.1097,
      "step": 5362,
      "task_loss": 0.18197688460350037
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795197576178865,
      "compression_loss": 0.0,
      "distillation_loss": 0.015219754539430141,
      "epoch": 5.09,
      "learning_rate": 8.642395291135527e-06,
      "loss": 0.014,
      "step": 5363,
      "task_loss": 0.003273945301771164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952022798257599,
      "compression_loss": 0.0,
      "distillation_loss": 0.03735918551683426,
      "epoch": 5.09,
      "learning_rate": 8.634338945000287e-06,
      "loss": 0.0442,
      "step": 5364,
      "task_loss": 0.10612202435731888
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795206980400384,
      "compression_loss": 0.0,
      "distillation_loss": 0.0819498747587204,
      "epoch": 5.09,
      "learning_rate": 8.626285571766204e-06,
      "loss": 0.1011,
      "step": 5365,
      "task_loss": 0.27321359515190125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952116779037407,
      "compression_loss": 0.0,
      "distillation_loss": 0.036407940089702606,
      "epoch": 5.1,
      "learning_rate": 8.618235172896203e-06,
      "loss": 0.0459,
      "step": 5366,
      "task_loss": 0.13110488653182983
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952163723368341,
      "compression_loss": 0.0,
      "distillation_loss": 0.03413936868309975,
      "epoch": 5.1,
      "learning_rate": 8.61018774985267e-06,
      "loss": 0.0388,
      "step": 5367,
      "task_loss": 0.08028659969568253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952210637006675,
      "compression_loss": 0.0,
      "distillation_loss": 0.026723839342594147,
      "epoch": 5.1,
      "learning_rate": 8.602143304097477e-06,
      "loss": 0.0248,
      "step": 5368,
      "task_loss": 0.007091682404279709
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952257519962448,
      "compression_loss": 0.0,
      "distillation_loss": 0.07885417342185974,
      "epoch": 5.1,
      "learning_rate": 8.594101837091922e-06,
      "loss": 0.0772,
      "step": 5369,
      "task_loss": 0.06278681010007858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952304372245697,
      "compression_loss": 0.0,
      "distillation_loss": 0.19364023208618164,
      "epoch": 5.1,
      "learning_rate": 8.586063350296775e-06,
      "loss": 0.1925,
      "step": 5370,
      "task_loss": 0.18187300860881805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952351193866457,
      "compression_loss": 0.0,
      "distillation_loss": 0.02647271379828453,
      "epoch": 5.1,
      "learning_rate": 8.578027845172269e-06,
      "loss": 0.0321,
      "step": 5371,
      "task_loss": 0.08231362700462341
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952397984834766,
      "compression_loss": 0.0,
      "distillation_loss": 0.04869615286588669,
      "epoch": 5.1,
      "learning_rate": 8.569995323178102e-06,
      "loss": 0.0519,
      "step": 5372,
      "task_loss": 0.08067583292722702
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952444745160662,
      "compression_loss": 0.0,
      "distillation_loss": 0.017870426177978516,
      "epoch": 5.1,
      "learning_rate": 8.561965785773413e-06,
      "loss": 0.0165,
      "step": 5373,
      "task_loss": 0.004511706531047821
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952491474854179,
      "compression_loss": 0.0,
      "distillation_loss": 0.010581446811556816,
      "epoch": 5.1,
      "learning_rate": 8.553939234416796e-06,
      "loss": 0.0097,
      "step": 5374,
      "task_loss": 0.0022147856652736664
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952538173925355,
      "compression_loss": 0.0,
      "distillation_loss": 0.04116586595773697,
      "epoch": 5.1,
      "learning_rate": 8.545915670566331e-06,
      "loss": 0.0379,
      "step": 5375,
      "task_loss": 0.008842799812555313
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952584842384228,
      "compression_loss": 0.0,
      "distillation_loss": 0.2459995001554489,
      "epoch": 5.11,
      "learning_rate": 8.53789509567953e-06,
      "loss": 0.2456,
      "step": 5376,
      "task_loss": 0.24173244833946228
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952631480240835,
      "compression_loss": 0.0,
      "distillation_loss": 0.09949977695941925,
      "epoch": 5.11,
      "learning_rate": 8.529877511213357e-06,
      "loss": 0.0938,
      "step": 5377,
      "task_loss": 0.04220582917332649
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795267808750521,
      "compression_loss": 0.0,
      "distillation_loss": 0.03112136758863926,
      "epoch": 5.11,
      "learning_rate": 8.521862918624265e-06,
      "loss": 0.0283,
      "step": 5378,
      "task_loss": 0.0033892393112182617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952724664187393,
      "compression_loss": 0.0,
      "distillation_loss": 0.08223070204257965,
      "epoch": 5.11,
      "learning_rate": 8.513851319368124e-06,
      "loss": 0.0853,
      "step": 5379,
      "task_loss": 0.11259977519512177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952771210297419,
      "compression_loss": 0.0,
      "distillation_loss": 0.07326701283454895,
      "epoch": 5.11,
      "learning_rate": 8.505842714900297e-06,
      "loss": 0.0902,
      "step": 5380,
      "task_loss": 0.24259977042675018
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952817725845325,
      "compression_loss": 0.0,
      "distillation_loss": 0.03324619308114052,
      "epoch": 5.11,
      "learning_rate": 8.497837106675571e-06,
      "loss": 0.0454,
      "step": 5381,
      "task_loss": 0.15457826852798462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952864210841147,
      "compression_loss": 0.0,
      "distillation_loss": 0.05445803701877594,
      "epoch": 5.11,
      "learning_rate": 8.489834496148217e-06,
      "loss": 0.0585,
      "step": 5382,
      "task_loss": 0.09474760293960571
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952910665294924,
      "compression_loss": 0.0,
      "distillation_loss": 0.05557304993271828,
      "epoch": 5.11,
      "learning_rate": 8.481834884771938e-06,
      "loss": 0.0507,
      "step": 5383,
      "task_loss": 0.006748411804437637
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7952957089216692,
      "compression_loss": 0.0,
      "distillation_loss": 0.09471587836742401,
      "epoch": 5.11,
      "learning_rate": 8.4738382739999e-06,
      "loss": 0.094,
      "step": 5384,
      "task_loss": 0.08728061616420746
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953003482616489,
      "compression_loss": 0.0,
      "distillation_loss": 0.10454539954662323,
      "epoch": 5.11,
      "learning_rate": 8.465844665284739e-06,
      "loss": 0.1136,
      "step": 5385,
      "task_loss": 0.19549594819545746
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953049845504347,
      "compression_loss": 0.0,
      "distillation_loss": 0.17727123200893402,
      "epoch": 5.11,
      "learning_rate": 8.45785406007852e-06,
      "loss": 0.1673,
      "step": 5386,
      "task_loss": 0.07721205800771713
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953096177890309,
      "compression_loss": 0.0,
      "distillation_loss": 0.02914445474743843,
      "epoch": 5.12,
      "learning_rate": 8.449866459832775e-06,
      "loss": 0.0333,
      "step": 5387,
      "task_loss": 0.0706995502114296
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953142479784407,
      "compression_loss": 0.0,
      "distillation_loss": 0.03571475297212601,
      "epoch": 5.12,
      "learning_rate": 8.44188186599849e-06,
      "loss": 0.0336,
      "step": 5388,
      "task_loss": 0.014664966613054276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953188751196681,
      "compression_loss": 0.0,
      "distillation_loss": 0.06461668014526367,
      "epoch": 5.12,
      "learning_rate": 8.433900280026118e-06,
      "loss": 0.0607,
      "step": 5389,
      "task_loss": 0.0250856913626194
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953234992137167,
      "compression_loss": 0.0,
      "distillation_loss": 0.09197506308555603,
      "epoch": 5.12,
      "learning_rate": 8.425921703365547e-06,
      "loss": 0.094,
      "step": 5390,
      "task_loss": 0.11221229285001755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953281202615902,
      "compression_loss": 0.0,
      "distillation_loss": 0.04114022105932236,
      "epoch": 5.12,
      "learning_rate": 8.417946137466107e-06,
      "loss": 0.0377,
      "step": 5391,
      "task_loss": 0.006361830979585648
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953327382642921,
      "compression_loss": 0.0,
      "distillation_loss": 0.11470133811235428,
      "epoch": 5.12,
      "learning_rate": 8.409973583776624e-06,
      "loss": 0.133,
      "step": 5392,
      "task_loss": 0.29758089780807495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953373532228262,
      "compression_loss": 0.0,
      "distillation_loss": 0.029153695330023766,
      "epoch": 5.12,
      "learning_rate": 8.402004043745335e-06,
      "loss": 0.0271,
      "step": 5393,
      "task_loss": 0.008810054510831833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953419651381962,
      "compression_loss": 0.0,
      "distillation_loss": 0.031133677810430527,
      "epoch": 5.12,
      "learning_rate": 8.39403751881994e-06,
      "loss": 0.0376,
      "step": 5394,
      "task_loss": 0.09556027501821518
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953465740114058,
      "compression_loss": 0.0,
      "distillation_loss": 0.08608759194612503,
      "epoch": 5.12,
      "learning_rate": 8.386074010447614e-06,
      "loss": 0.0831,
      "step": 5395,
      "task_loss": 0.05571907386183739
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953511798434587,
      "compression_loss": 0.0,
      "distillation_loss": 0.028766460716724396,
      "epoch": 5.12,
      "learning_rate": 8.378113520074949e-06,
      "loss": 0.0262,
      "step": 5396,
      "task_loss": 0.003098204731941223
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953557826353584,
      "compression_loss": 0.0,
      "distillation_loss": 0.09285937994718552,
      "epoch": 5.13,
      "learning_rate": 8.370156049148022e-06,
      "loss": 0.0864,
      "step": 5397,
      "task_loss": 0.027815401554107666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953603823881089,
      "compression_loss": 0.0,
      "distillation_loss": 0.07426916807889938,
      "epoch": 5.13,
      "learning_rate": 8.362201599112332e-06,
      "loss": 0.0757,
      "step": 5398,
      "task_loss": 0.08855120837688446
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953649791027136,
      "compression_loss": 0.0,
      "distillation_loss": 0.026602642610669136,
      "epoch": 5.13,
      "learning_rate": 8.354250171412859e-06,
      "loss": 0.0299,
      "step": 5399,
      "task_loss": 0.0595041885972023
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953695727801764,
      "compression_loss": 0.0,
      "distillation_loss": 0.10936440527439117,
      "epoch": 5.13,
      "learning_rate": 8.346301767494008e-06,
      "loss": 0.1146,
      "step": 5400,
      "task_loss": 0.16161611676216125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953741634215007,
      "compression_loss": 0.0,
      "distillation_loss": 0.021258335560560226,
      "epoch": 5.13,
      "learning_rate": 8.338356388799637e-06,
      "loss": 0.0326,
      "step": 5401,
      "task_loss": 0.1344192773103714
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953787510276905,
      "compression_loss": 0.0,
      "distillation_loss": 0.04099886491894722,
      "epoch": 5.13,
      "learning_rate": 8.330414036773082e-06,
      "loss": 0.0412,
      "step": 5402,
      "task_loss": 0.04298762232065201
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953833355997493,
      "compression_loss": 0.0,
      "distillation_loss": 0.07252480089664459,
      "epoch": 5.13,
      "learning_rate": 8.322474712857095e-06,
      "loss": 0.0676,
      "step": 5403,
      "task_loss": 0.023213542997837067
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953879171386807,
      "compression_loss": 0.0,
      "distillation_loss": 0.023728784173727036,
      "epoch": 5.13,
      "learning_rate": 8.314538418493892e-06,
      "loss": 0.0218,
      "step": 5404,
      "task_loss": 0.004381675273180008
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953924956454886,
      "compression_loss": 0.0,
      "distillation_loss": 0.05856727808713913,
      "epoch": 5.13,
      "learning_rate": 8.306605155125141e-06,
      "loss": 0.0642,
      "step": 5405,
      "task_loss": 0.11514223366975784
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7953970711211765,
      "compression_loss": 0.0,
      "distillation_loss": 0.022047195583581924,
      "epoch": 5.13,
      "learning_rate": 8.298674924191968e-06,
      "loss": 0.0206,
      "step": 5406,
      "task_loss": 0.0073068346828222275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954016435667483,
      "compression_loss": 0.0,
      "distillation_loss": 0.02194095030426979,
      "epoch": 5.13,
      "learning_rate": 8.29074772713493e-06,
      "loss": 0.0201,
      "step": 5407,
      "task_loss": 0.0031332336366176605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954062129832075,
      "compression_loss": 0.0,
      "distillation_loss": 0.03532099723815918,
      "epoch": 5.14,
      "learning_rate": 8.282823565394032e-06,
      "loss": 0.0414,
      "step": 5408,
      "task_loss": 0.09565601497888565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954107793715578,
      "compression_loss": 0.0,
      "distillation_loss": 0.12141124904155731,
      "epoch": 5.14,
      "learning_rate": 8.274902440408755e-06,
      "loss": 0.125,
      "step": 5409,
      "task_loss": 0.1574663519859314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954153427328029,
      "compression_loss": 0.0,
      "distillation_loss": 0.0844094529747963,
      "epoch": 5.14,
      "learning_rate": 8.266984353617995e-06,
      "loss": 0.092,
      "step": 5410,
      "task_loss": 0.16067147254943848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954199030679464,
      "compression_loss": 0.0,
      "distillation_loss": 0.07113252580165863,
      "epoch": 5.14,
      "learning_rate": 8.259069306460107e-06,
      "loss": 0.0768,
      "step": 5411,
      "task_loss": 0.12745647132396698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954244603779922,
      "compression_loss": 0.0,
      "distillation_loss": 0.12387774884700775,
      "epoch": 5.14,
      "learning_rate": 8.251157300372913e-06,
      "loss": 0.1206,
      "step": 5412,
      "task_loss": 0.09093039482831955
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954290146639438,
      "compression_loss": 0.0,
      "distillation_loss": 0.12096986174583435,
      "epoch": 5.14,
      "learning_rate": 8.243248336793658e-06,
      "loss": 0.1201,
      "step": 5413,
      "task_loss": 0.11250494420528412
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954335659268049,
      "compression_loss": 0.0,
      "distillation_loss": 0.09077440202236176,
      "epoch": 5.14,
      "learning_rate": 8.235342417159036e-06,
      "loss": 0.0899,
      "step": 5414,
      "task_loss": 0.08181983232498169
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954381141675794,
      "compression_loss": 0.0,
      "distillation_loss": 0.08009003102779388,
      "epoch": 5.14,
      "learning_rate": 8.227439542905205e-06,
      "loss": 0.0828,
      "step": 5415,
      "task_loss": 0.10733769834041595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954426593872705,
      "compression_loss": 0.0,
      "distillation_loss": 0.05436462163925171,
      "epoch": 5.14,
      "learning_rate": 8.219539715467766e-06,
      "loss": 0.0544,
      "step": 5416,
      "task_loss": 0.05464364215731621
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954472015868824,
      "compression_loss": 0.0,
      "distillation_loss": 0.029182853177189827,
      "epoch": 5.14,
      "learning_rate": 8.211642936281752e-06,
      "loss": 0.0272,
      "step": 5417,
      "task_loss": 0.00951925665140152
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954517407674185,
      "compression_loss": 0.0,
      "distillation_loss": 0.027566958218812943,
      "epoch": 5.15,
      "learning_rate": 8.203749206781647e-06,
      "loss": 0.0264,
      "step": 5418,
      "task_loss": 0.015928007662296295
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954562769298826,
      "compression_loss": 0.0,
      "distillation_loss": 0.17319881916046143,
      "epoch": 5.15,
      "learning_rate": 8.195858528401396e-06,
      "loss": 0.1678,
      "step": 5419,
      "task_loss": 0.11959183216094971
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954608100752782,
      "compression_loss": 0.0,
      "distillation_loss": 0.012766760773956776,
      "epoch": 5.15,
      "learning_rate": 8.187970902574371e-06,
      "loss": 0.0118,
      "step": 5420,
      "task_loss": 0.002884853631258011
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954653402046092,
      "compression_loss": 0.0,
      "distillation_loss": 0.1071738451719284,
      "epoch": 5.15,
      "learning_rate": 8.180086330733403e-06,
      "loss": 0.1129,
      "step": 5421,
      "task_loss": 0.16459250450134277
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954698673188793,
      "compression_loss": 0.0,
      "distillation_loss": 0.06729540228843689,
      "epoch": 5.15,
      "learning_rate": 8.172204814310742e-06,
      "loss": 0.0673,
      "step": 5422,
      "task_loss": 0.06722187995910645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954743914190919,
      "compression_loss": 0.0,
      "distillation_loss": 0.041276026517152786,
      "epoch": 5.15,
      "learning_rate": 8.164326354738134e-06,
      "loss": 0.0452,
      "step": 5423,
      "task_loss": 0.0802101194858551
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954789125062509,
      "compression_loss": 0.0,
      "distillation_loss": 0.04235687106847763,
      "epoch": 5.15,
      "learning_rate": 8.156450953446725e-06,
      "loss": 0.0563,
      "step": 5424,
      "task_loss": 0.18132589757442474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954834305813601,
      "compression_loss": 0.0,
      "distillation_loss": 0.041152555495500565,
      "epoch": 5.15,
      "learning_rate": 8.148578611867114e-06,
      "loss": 0.051,
      "step": 5425,
      "task_loss": 0.13986477255821228
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954879456454228,
      "compression_loss": 0.0,
      "distillation_loss": 0.034636110067367554,
      "epoch": 5.15,
      "learning_rate": 8.14070933142936e-06,
      "loss": 0.0315,
      "step": 5426,
      "task_loss": 0.003546604886651039
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795492457699443,
      "compression_loss": 0.0,
      "distillation_loss": 0.040368109941482544,
      "epoch": 5.15,
      "learning_rate": 8.132843113562954e-06,
      "loss": 0.0446,
      "step": 5427,
      "task_loss": 0.08240634202957153
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7954969667444243,
      "compression_loss": 0.0,
      "distillation_loss": 0.028532054275274277,
      "epoch": 5.15,
      "learning_rate": 8.12497995969683e-06,
      "loss": 0.0265,
      "step": 5428,
      "task_loss": 0.008003626018762589
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955014727813703,
      "compression_loss": 0.0,
      "distillation_loss": 0.019893258810043335,
      "epoch": 5.16,
      "learning_rate": 8.11711987125936e-06,
      "loss": 0.019,
      "step": 5429,
      "task_loss": 0.01055414229631424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955059758112848,
      "compression_loss": 0.0,
      "distillation_loss": 0.06559453159570694,
      "epoch": 5.16,
      "learning_rate": 8.109262849678378e-06,
      "loss": 0.066,
      "step": 5430,
      "task_loss": 0.06944157928228378
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955104758351714,
      "compression_loss": 0.0,
      "distillation_loss": 0.07780923694372177,
      "epoch": 5.16,
      "learning_rate": 8.101408896381141e-06,
      "loss": 0.0707,
      "step": 5431,
      "task_loss": 0.006445575505495071
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955149728540338,
      "compression_loss": 0.0,
      "distillation_loss": 0.049274250864982605,
      "epoch": 5.16,
      "learning_rate": 8.093558012794363e-06,
      "loss": 0.0533,
      "step": 5432,
      "task_loss": 0.08973647654056549
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955194668688758,
      "compression_loss": 0.0,
      "distillation_loss": 0.038362838327884674,
      "epoch": 5.16,
      "learning_rate": 8.085710200344202e-06,
      "loss": 0.0464,
      "step": 5433,
      "task_loss": 0.1189565435051918
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955239578807008,
      "compression_loss": 0.0,
      "distillation_loss": 0.14410632848739624,
      "epoch": 5.16,
      "learning_rate": 8.077865460456241e-06,
      "loss": 0.1396,
      "step": 5434,
      "task_loss": 0.09874996542930603
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955284458905129,
      "compression_loss": 0.0,
      "distillation_loss": 0.019637832418084145,
      "epoch": 5.16,
      "learning_rate": 8.070023794555521e-06,
      "loss": 0.0285,
      "step": 5435,
      "task_loss": 0.1084957867860794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955329308993153,
      "compression_loss": 0.0,
      "distillation_loss": 0.022479506209492683,
      "epoch": 5.16,
      "learning_rate": 8.062185204066503e-06,
      "loss": 0.0287,
      "step": 5436,
      "task_loss": 0.08433564752340317
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955374129081121,
      "compression_loss": 0.0,
      "distillation_loss": 0.018630899488925934,
      "epoch": 5.16,
      "learning_rate": 8.054349690413124e-06,
      "loss": 0.0186,
      "step": 5437,
      "task_loss": 0.018421385437250137
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955418919179067,
      "compression_loss": 0.0,
      "distillation_loss": 0.1612919569015503,
      "epoch": 5.16,
      "learning_rate": 8.046517255018737e-06,
      "loss": 0.1564,
      "step": 5438,
      "task_loss": 0.11284254491329193
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955463679297029,
      "compression_loss": 0.0,
      "distillation_loss": 0.02024524286389351,
      "epoch": 5.17,
      "learning_rate": 8.03868789930613e-06,
      "loss": 0.0189,
      "step": 5439,
      "task_loss": 0.006963614374399185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955508409445043,
      "compression_loss": 0.0,
      "distillation_loss": 0.08255909383296967,
      "epoch": 5.17,
      "learning_rate": 8.030861624697552e-06,
      "loss": 0.0821,
      "step": 5440,
      "task_loss": 0.07832753658294678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955553109633149,
      "compression_loss": 0.0,
      "distillation_loss": 0.03200299292802811,
      "epoch": 5.17,
      "learning_rate": 8.023038432614693e-06,
      "loss": 0.0304,
      "step": 5441,
      "task_loss": 0.01600661687552929
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795559777987138,
      "compression_loss": 0.0,
      "distillation_loss": 0.044372882694005966,
      "epoch": 5.17,
      "learning_rate": 8.015218324478666e-06,
      "loss": 0.0583,
      "step": 5442,
      "task_loss": 0.18402326107025146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955642420169773,
      "compression_loss": 0.0,
      "distillation_loss": 0.03316294774413109,
      "epoch": 5.17,
      "learning_rate": 8.007401301710022e-06,
      "loss": 0.0323,
      "step": 5443,
      "task_loss": 0.02459472045302391
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955687030538368,
      "compression_loss": 0.0,
      "distillation_loss": 0.04046769440174103,
      "epoch": 5.17,
      "learning_rate": 7.999587365728776e-06,
      "loss": 0.0457,
      "step": 5444,
      "task_loss": 0.09298967570066452
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955731610987198,
      "compression_loss": 0.0,
      "distillation_loss": 0.03207171708345413,
      "epoch": 5.17,
      "learning_rate": 7.991776517954359e-06,
      "loss": 0.0461,
      "step": 5445,
      "task_loss": 0.17218343913555145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955776161526302,
      "compression_loss": 0.0,
      "distillation_loss": 0.05809757858514786,
      "epoch": 5.17,
      "learning_rate": 7.983968759805641e-06,
      "loss": 0.0545,
      "step": 5446,
      "task_loss": 0.022373545914888382
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955820682165717,
      "compression_loss": 0.0,
      "distillation_loss": 0.03011673502624035,
      "epoch": 5.17,
      "learning_rate": 7.976164092700958e-06,
      "loss": 0.0321,
      "step": 5447,
      "task_loss": 0.050249215215444565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795586517291548,
      "compression_loss": 0.0,
      "distillation_loss": 0.08846057951450348,
      "epoch": 5.17,
      "learning_rate": 7.968362518058048e-06,
      "loss": 0.085,
      "step": 5448,
      "task_loss": 0.05429108440876007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955909633785625,
      "compression_loss": 0.0,
      "distillation_loss": 0.2117672562599182,
      "epoch": 5.17,
      "learning_rate": 7.960564037294118e-06,
      "loss": 0.2183,
      "step": 5449,
      "task_loss": 0.276851087808609
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955954064786193,
      "compression_loss": 0.0,
      "distillation_loss": 0.06762327998876572,
      "epoch": 5.18,
      "learning_rate": 7.952768651825784e-06,
      "loss": 0.0656,
      "step": 5450,
      "task_loss": 0.04782693088054657
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7955998465927218,
      "compression_loss": 0.0,
      "distillation_loss": 0.0332944318652153,
      "epoch": 5.18,
      "learning_rate": 7.944976363069137e-06,
      "loss": 0.0315,
      "step": 5451,
      "task_loss": 0.015227574855089188
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956042837218736,
      "compression_loss": 0.0,
      "distillation_loss": 0.15575581789016724,
      "epoch": 5.18,
      "learning_rate": 7.937187172439669e-06,
      "loss": 0.1656,
      "step": 5452,
      "task_loss": 0.25407275557518005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956087178670788,
      "compression_loss": 0.0,
      "distillation_loss": 0.04357504844665527,
      "epoch": 5.18,
      "learning_rate": 7.929401081352319e-06,
      "loss": 0.0438,
      "step": 5453,
      "task_loss": 0.045449040830135345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956131490293407,
      "compression_loss": 0.0,
      "distillation_loss": 0.0458906814455986,
      "epoch": 5.18,
      "learning_rate": 7.921618091221484e-06,
      "loss": 0.047,
      "step": 5454,
      "task_loss": 0.05692676454782486
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956175772096631,
      "compression_loss": 0.0,
      "distillation_loss": 0.023164385929703712,
      "epoch": 5.18,
      "learning_rate": 7.91383820346097e-06,
      "loss": 0.0319,
      "step": 5455,
      "task_loss": 0.1104351356625557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956220024090497,
      "compression_loss": 0.0,
      "distillation_loss": 0.04113226756453514,
      "epoch": 5.18,
      "learning_rate": 7.906061419484028e-06,
      "loss": 0.0444,
      "step": 5456,
      "task_loss": 0.07416713237762451
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956264246285041,
      "compression_loss": 0.0,
      "distillation_loss": 0.043819695711135864,
      "epoch": 5.18,
      "learning_rate": 7.898287740703356e-06,
      "loss": 0.0427,
      "step": 5457,
      "task_loss": 0.03260737285017967
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79563084386903,
      "compression_loss": 0.0,
      "distillation_loss": 0.028930241242051125,
      "epoch": 5.18,
      "learning_rate": 7.890517168531086e-06,
      "loss": 0.0365,
      "step": 5458,
      "task_loss": 0.1042308360338211
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956352601316313,
      "compression_loss": 0.0,
      "distillation_loss": 0.20450790226459503,
      "epoch": 5.18,
      "learning_rate": 7.882749704378773e-06,
      "loss": 0.1955,
      "step": 5459,
      "task_loss": 0.11405302584171295
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956396734173115,
      "compression_loss": 0.0,
      "distillation_loss": 0.048243820667266846,
      "epoch": 5.19,
      "learning_rate": 7.874985349657405e-06,
      "loss": 0.0511,
      "step": 5460,
      "task_loss": 0.07671748101711273
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956440837270742,
      "compression_loss": 0.0,
      "distillation_loss": 0.012140437960624695,
      "epoch": 5.19,
      "learning_rate": 7.86722410577743e-06,
      "loss": 0.0113,
      "step": 5461,
      "task_loss": 0.003828030079603195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956484910619231,
      "compression_loss": 0.0,
      "distillation_loss": 0.02375885099172592,
      "epoch": 5.19,
      "learning_rate": 7.85946597414871e-06,
      "loss": 0.0221,
      "step": 5462,
      "task_loss": 0.006828447803854942
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956528954228621,
      "compression_loss": 0.0,
      "distillation_loss": 0.07895895093679428,
      "epoch": 5.19,
      "learning_rate": 7.85171095618054e-06,
      "loss": 0.0807,
      "step": 5463,
      "task_loss": 0.0966450572013855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956572968108947,
      "compression_loss": 0.0,
      "distillation_loss": 0.025494446977972984,
      "epoch": 5.19,
      "learning_rate": 7.843959053281663e-06,
      "loss": 0.0237,
      "step": 5464,
      "task_loss": 0.007278071716427803
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956616952270247,
      "compression_loss": 0.0,
      "distillation_loss": 0.020559193566441536,
      "epoch": 5.19,
      "learning_rate": 7.836210266860253e-06,
      "loss": 0.0241,
      "step": 5465,
      "task_loss": 0.05645029619336128
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956660906722556,
      "compression_loss": 0.0,
      "distillation_loss": 0.0954541563987732,
      "epoch": 5.19,
      "learning_rate": 7.8284645983239e-06,
      "loss": 0.0991,
      "step": 5466,
      "task_loss": 0.1314251720905304
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956704831475913,
      "compression_loss": 0.0,
      "distillation_loss": 0.047063831239938736,
      "epoch": 5.19,
      "learning_rate": 7.820722049079653e-06,
      "loss": 0.0666,
      "step": 5467,
      "task_loss": 0.24207653105258942
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956748726540352,
      "compression_loss": 0.0,
      "distillation_loss": 0.04180522635579109,
      "epoch": 5.19,
      "learning_rate": 7.812982620533993e-06,
      "loss": 0.0509,
      "step": 5468,
      "task_loss": 0.13276955485343933
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956792591925913,
      "compression_loss": 0.0,
      "distillation_loss": 0.01592065393924713,
      "epoch": 5.19,
      "learning_rate": 7.805246314092809e-06,
      "loss": 0.0236,
      "step": 5469,
      "task_loss": 0.09271353483200073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956836427642631,
      "compression_loss": 0.0,
      "distillation_loss": 0.032307371497154236,
      "epoch": 5.19,
      "learning_rate": 7.797513131161437e-06,
      "loss": 0.0293,
      "step": 5470,
      "task_loss": 0.002665068954229355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956880233700542,
      "compression_loss": 0.0,
      "distillation_loss": 0.042782217264175415,
      "epoch": 5.2,
      "learning_rate": 7.78978307314466e-06,
      "loss": 0.0452,
      "step": 5471,
      "task_loss": 0.06668760627508163
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956924010109685,
      "compression_loss": 0.0,
      "distillation_loss": 0.02939591184258461,
      "epoch": 5.2,
      "learning_rate": 7.782056141446673e-06,
      "loss": 0.0269,
      "step": 5472,
      "task_loss": 0.003960005939006805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7956967756880096,
      "compression_loss": 0.0,
      "distillation_loss": 0.13202489912509918,
      "epoch": 5.2,
      "learning_rate": 7.774332337471101e-06,
      "loss": 0.136,
      "step": 5473,
      "task_loss": 0.17181508243083954
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957011474021811,
      "compression_loss": 0.0,
      "distillation_loss": 0.02327493578195572,
      "epoch": 5.2,
      "learning_rate": 7.766611662621023e-06,
      "loss": 0.0219,
      "step": 5474,
      "task_loss": 0.009945593774318695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957055161544868,
      "compression_loss": 0.0,
      "distillation_loss": 0.05180184915661812,
      "epoch": 5.2,
      "learning_rate": 7.75889411829894e-06,
      "loss": 0.0526,
      "step": 5475,
      "task_loss": 0.05985014885663986
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957098819459303,
      "compression_loss": 0.0,
      "distillation_loss": 0.09386663138866425,
      "epoch": 5.2,
      "learning_rate": 7.751179705906774e-06,
      "loss": 0.0984,
      "step": 5476,
      "task_loss": 0.13911299407482147
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957142447775153,
      "compression_loss": 0.0,
      "distillation_loss": 0.060429397970438004,
      "epoch": 5.2,
      "learning_rate": 7.74346842684588e-06,
      "loss": 0.0566,
      "step": 5477,
      "task_loss": 0.022498732432723045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957186046502455,
      "compression_loss": 0.0,
      "distillation_loss": 0.02577110007405281,
      "epoch": 5.2,
      "learning_rate": 7.73576028251706e-06,
      "loss": 0.0237,
      "step": 5478,
      "task_loss": 0.005201183259487152
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957229615651246,
      "compression_loss": 0.0,
      "distillation_loss": 0.04601401090621948,
      "epoch": 5.2,
      "learning_rate": 7.728055274320527e-06,
      "loss": 0.0437,
      "step": 5479,
      "task_loss": 0.02243119105696678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957273155231561,
      "compression_loss": 0.0,
      "distillation_loss": 0.11642314493656158,
      "epoch": 5.2,
      "learning_rate": 7.720353403655934e-06,
      "loss": 0.1139,
      "step": 5480,
      "task_loss": 0.09131792187690735
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795731666525344,
      "compression_loss": 0.0,
      "distillation_loss": 0.06120690703392029,
      "epoch": 5.21,
      "learning_rate": 7.712654671922371e-06,
      "loss": 0.0697,
      "step": 5481,
      "task_loss": 0.14600446820259094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957360145726917,
      "compression_loss": 0.0,
      "distillation_loss": 0.014658894389867783,
      "epoch": 5.21,
      "learning_rate": 7.704959080518343e-06,
      "loss": 0.0295,
      "step": 5482,
      "task_loss": 0.16297921538352966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957403596662032,
      "compression_loss": 0.0,
      "distillation_loss": 0.05037935450673103,
      "epoch": 5.21,
      "learning_rate": 7.697266630841784e-06,
      "loss": 0.0564,
      "step": 5483,
      "task_loss": 0.1106264591217041
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957447018068818,
      "compression_loss": 0.0,
      "distillation_loss": 0.02987675368785858,
      "epoch": 5.21,
      "learning_rate": 7.689577324290073e-06,
      "loss": 0.0273,
      "step": 5484,
      "task_loss": 0.004036994650959969
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957490409957314,
      "compression_loss": 0.0,
      "distillation_loss": 0.06058541685342789,
      "epoch": 5.21,
      "learning_rate": 7.681891162260015e-06,
      "loss": 0.061,
      "step": 5485,
      "task_loss": 0.06461737304925919
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957533772337557,
      "compression_loss": 0.0,
      "distillation_loss": 0.024111486971378326,
      "epoch": 5.21,
      "learning_rate": 7.674208146147834e-06,
      "loss": 0.0267,
      "step": 5486,
      "task_loss": 0.04959327355027199
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957577105219583,
      "compression_loss": 0.0,
      "distillation_loss": 0.058419279754161835,
      "epoch": 5.21,
      "learning_rate": 7.666528277349177e-06,
      "loss": 0.0589,
      "step": 5487,
      "task_loss": 0.06302666664123535
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957620408613428,
      "compression_loss": 0.0,
      "distillation_loss": 0.04451620578765869,
      "epoch": 5.21,
      "learning_rate": 7.658851557259144e-06,
      "loss": 0.0585,
      "step": 5488,
      "task_loss": 0.18458762764930725
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957663682529131,
      "compression_loss": 0.0,
      "distillation_loss": 0.047050587832927704,
      "epoch": 5.21,
      "learning_rate": 7.651177987272243e-06,
      "loss": 0.0492,
      "step": 5489,
      "task_loss": 0.06901868432760239
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957706926976729,
      "compression_loss": 0.0,
      "distillation_loss": 0.03578869253396988,
      "epoch": 5.21,
      "learning_rate": 7.643507568782407e-06,
      "loss": 0.033,
      "step": 5490,
      "task_loss": 0.007528049871325493
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957750141966256,
      "compression_loss": 0.0,
      "distillation_loss": 0.04843487590551376,
      "epoch": 5.21,
      "learning_rate": 7.635840303183018e-06,
      "loss": 0.0525,
      "step": 5491,
      "task_loss": 0.08893577009439468
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957793327507751,
      "compression_loss": 0.0,
      "distillation_loss": 0.014655525796115398,
      "epoch": 5.22,
      "learning_rate": 7.628176191866853e-06,
      "loss": 0.0138,
      "step": 5492,
      "task_loss": 0.005986766889691353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957836483611249,
      "compression_loss": 0.0,
      "distillation_loss": 0.036418359726667404,
      "epoch": 5.22,
      "learning_rate": 7.6205152362261586e-06,
      "loss": 0.0488,
      "step": 5493,
      "task_loss": 0.1600310057401657
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957879610286789,
      "compression_loss": 0.0,
      "distillation_loss": 0.031819455325603485,
      "epoch": 5.22,
      "learning_rate": 7.612857437652563e-06,
      "loss": 0.0348,
      "step": 5494,
      "task_loss": 0.06123275309801102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957922707544407,
      "compression_loss": 0.0,
      "distillation_loss": 0.0890742614865303,
      "epoch": 5.22,
      "learning_rate": 7.605202797537156e-06,
      "loss": 0.0832,
      "step": 5495,
      "task_loss": 0.03046075440943241
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7957965775394139,
      "compression_loss": 0.0,
      "distillation_loss": 0.06095193326473236,
      "epoch": 5.22,
      "learning_rate": 7.5975513172704375e-06,
      "loss": 0.0706,
      "step": 5496,
      "task_loss": 0.15771833062171936
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958008813846024,
      "compression_loss": 0.0,
      "distillation_loss": 0.04849329590797424,
      "epoch": 5.22,
      "learning_rate": 7.589902998242327e-06,
      "loss": 0.0612,
      "step": 5497,
      "task_loss": 0.17582949995994568
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958051822910096,
      "compression_loss": 0.0,
      "distillation_loss": 0.08853879570960999,
      "epoch": 5.22,
      "learning_rate": 7.582257841842191e-06,
      "loss": 0.0895,
      "step": 5498,
      "task_loss": 0.09831196069717407
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958094802596393,
      "compression_loss": 0.0,
      "distillation_loss": 0.029421133920550346,
      "epoch": 5.22,
      "learning_rate": 7.574615849458805e-06,
      "loss": 0.0488,
      "step": 5499,
      "task_loss": 0.22320178151130676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958137752914952,
      "compression_loss": 0.0,
      "distillation_loss": 0.10746181011199951,
      "epoch": 5.22,
      "learning_rate": 7.56697702248036e-06,
      "loss": 0.1032,
      "step": 5500,
      "task_loss": 0.06493166834115982
    },
    {
      "epoch": 5.22,
      "eval_accuracy": 0.8795871559633027,
      "eval_loss": 0.5020073056221008,
      "eval_runtime": 18.0294,
      "eval_samples_per_second": 48.365,
      "eval_steps_per_second": 6.046,
      "step": 5500
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958180673875811,
      "compression_loss": 0.0,
      "distillation_loss": 0.07600829005241394,
      "epoch": 5.22,
      "learning_rate": 7.5593413622945e-06,
      "loss": 0.0773,
      "step": 5501,
      "task_loss": 0.08857627213001251
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958223565489004,
      "compression_loss": 0.0,
      "distillation_loss": 0.06297820806503296,
      "epoch": 5.23,
      "learning_rate": 7.551708870288282e-06,
      "loss": 0.0582,
      "step": 5502,
      "task_loss": 0.0155994463711977
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958266427764571,
      "compression_loss": 0.0,
      "distillation_loss": 0.045012425631284714,
      "epoch": 5.23,
      "learning_rate": 7.5440795478481815e-06,
      "loss": 0.0409,
      "step": 5503,
      "task_loss": 0.0038008801639080048
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958309260712546,
      "compression_loss": 0.0,
      "distillation_loss": 0.02455669268965721,
      "epoch": 5.23,
      "learning_rate": 7.536453396360091e-06,
      "loss": 0.025,
      "step": 5504,
      "task_loss": 0.0289287306368351
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958352064342967,
      "compression_loss": 0.0,
      "distillation_loss": 0.082015261054039,
      "epoch": 5.23,
      "learning_rate": 7.528830417209356e-06,
      "loss": 0.0851,
      "step": 5505,
      "task_loss": 0.11327110230922699
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958394838665872,
      "compression_loss": 0.0,
      "distillation_loss": 0.025041064247488976,
      "epoch": 5.23,
      "learning_rate": 7.521210611780716e-06,
      "loss": 0.024,
      "step": 5506,
      "task_loss": 0.014456404373049736
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958437583691296,
      "compression_loss": 0.0,
      "distillation_loss": 0.1386338621377945,
      "epoch": 5.23,
      "learning_rate": 7.513593981458347e-06,
      "loss": 0.1412,
      "step": 5507,
      "task_loss": 0.1647292524576187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958480299429276,
      "compression_loss": 0.0,
      "distillation_loss": 0.11838963627815247,
      "epoch": 5.23,
      "learning_rate": 7.505980527625841e-06,
      "loss": 0.1125,
      "step": 5508,
      "task_loss": 0.0594179704785347
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795852298588985,
      "compression_loss": 0.0,
      "distillation_loss": 0.035721536725759506,
      "epoch": 5.23,
      "learning_rate": 7.498370251666223e-06,
      "loss": 0.0433,
      "step": 5509,
      "task_loss": 0.1113739162683487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958565643083054,
      "compression_loss": 0.0,
      "distillation_loss": 0.04196429252624512,
      "epoch": 5.23,
      "learning_rate": 7.490763154961944e-06,
      "loss": 0.0402,
      "step": 5510,
      "task_loss": 0.02450537495315075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958608271018925,
      "compression_loss": 0.0,
      "distillation_loss": 0.15733975172042847,
      "epoch": 5.23,
      "learning_rate": 7.483159238894858e-06,
      "loss": 0.156,
      "step": 5511,
      "task_loss": 0.14443063735961914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79586508697075,
      "compression_loss": 0.0,
      "distillation_loss": 0.021348509937524796,
      "epoch": 5.23,
      "learning_rate": 7.475558504846264e-06,
      "loss": 0.0288,
      "step": 5512,
      "task_loss": 0.09601381421089172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958693439158816,
      "compression_loss": 0.0,
      "distillation_loss": 0.03914246708154678,
      "epoch": 5.24,
      "learning_rate": 7.467960954196867e-06,
      "loss": 0.0441,
      "step": 5513,
      "task_loss": 0.08865661174058914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958735979382908,
      "compression_loss": 0.0,
      "distillation_loss": 0.1477883756160736,
      "epoch": 5.24,
      "learning_rate": 7.460366588326797e-06,
      "loss": 0.1582,
      "step": 5514,
      "task_loss": 0.25182464718818665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958778490389816,
      "compression_loss": 0.0,
      "distillation_loss": 0.041810937225818634,
      "epoch": 5.24,
      "learning_rate": 7.452775408615603e-06,
      "loss": 0.0481,
      "step": 5515,
      "task_loss": 0.10518115013837814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958820972189574,
      "compression_loss": 0.0,
      "distillation_loss": 0.01617216318845749,
      "epoch": 5.24,
      "learning_rate": 7.44518741644227e-06,
      "loss": 0.0226,
      "step": 5516,
      "task_loss": 0.07998733222484589
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958863424792221,
      "compression_loss": 0.0,
      "distillation_loss": 0.05367042124271393,
      "epoch": 5.24,
      "learning_rate": 7.4376026131851875e-06,
      "loss": 0.0558,
      "step": 5517,
      "task_loss": 0.07507097721099854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958905848207791,
      "compression_loss": 0.0,
      "distillation_loss": 0.07538196444511414,
      "epoch": 5.24,
      "learning_rate": 7.430021000222156e-06,
      "loss": 0.0736,
      "step": 5518,
      "task_loss": 0.05794493108987808
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958948242446324,
      "compression_loss": 0.0,
      "distillation_loss": 0.09253949671983719,
      "epoch": 5.24,
      "learning_rate": 7.422442578930444e-06,
      "loss": 0.0853,
      "step": 5519,
      "task_loss": 0.019849814474582672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7958990607517855,
      "compression_loss": 0.0,
      "distillation_loss": 0.016655316576361656,
      "epoch": 5.24,
      "learning_rate": 7.414867350686694e-06,
      "loss": 0.0227,
      "step": 5520,
      "task_loss": 0.07758677005767822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959032943432421,
      "compression_loss": 0.0,
      "distillation_loss": 0.06732715666294098,
      "epoch": 5.24,
      "learning_rate": 7.407295316866977e-06,
      "loss": 0.0669,
      "step": 5521,
      "task_loss": 0.06263066828250885
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795907525020006,
      "compression_loss": 0.0,
      "distillation_loss": 0.10630275309085846,
      "epoch": 5.24,
      "learning_rate": 7.399726478846788e-06,
      "loss": 0.1214,
      "step": 5522,
      "task_loss": 0.2570263147354126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959117527830806,
      "compression_loss": 0.0,
      "distillation_loss": 0.05409185588359833,
      "epoch": 5.25,
      "learning_rate": 7.392160838001055e-06,
      "loss": 0.0691,
      "step": 5523,
      "task_loss": 0.20388929545879364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959159776334699,
      "compression_loss": 0.0,
      "distillation_loss": 0.04524946212768555,
      "epoch": 5.25,
      "learning_rate": 7.384598395704106e-06,
      "loss": 0.0479,
      "step": 5524,
      "task_loss": 0.0716903880238533
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959201995721774,
      "compression_loss": 0.0,
      "distillation_loss": 0.2098562866449356,
      "epoch": 5.25,
      "learning_rate": 7.377039153329687e-06,
      "loss": 0.2018,
      "step": 5525,
      "task_loss": 0.12896330654621124
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959244186002068,
      "compression_loss": 0.0,
      "distillation_loss": 0.037853680551052094,
      "epoch": 5.25,
      "learning_rate": 7.36948311225098e-06,
      "loss": 0.0462,
      "step": 5526,
      "task_loss": 0.12155872583389282
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959286347185619,
      "compression_loss": 0.0,
      "distillation_loss": 0.018217723816633224,
      "epoch": 5.25,
      "learning_rate": 7.361930273840581e-06,
      "loss": 0.0254,
      "step": 5527,
      "task_loss": 0.09013350307941437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959328479282463,
      "compression_loss": 0.0,
      "distillation_loss": 0.07805530726909637,
      "epoch": 5.25,
      "learning_rate": 7.3543806394704955e-06,
      "loss": 0.0762,
      "step": 5528,
      "task_loss": 0.05900496616959572
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959370582302636,
      "compression_loss": 0.0,
      "distillation_loss": 0.07341839373111725,
      "epoch": 5.25,
      "learning_rate": 7.346834210512138e-06,
      "loss": 0.0724,
      "step": 5529,
      "task_loss": 0.06279435008764267
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959412656256175,
      "compression_loss": 0.0,
      "distillation_loss": 0.02832254022359848,
      "epoch": 5.25,
      "learning_rate": 7.3392909883363755e-06,
      "loss": 0.0378,
      "step": 5530,
      "task_loss": 0.12311002612113953
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959454701153119,
      "compression_loss": 0.0,
      "distillation_loss": 0.04221939295530319,
      "epoch": 5.25,
      "learning_rate": 7.331750974313459e-06,
      "loss": 0.0418,
      "step": 5531,
      "task_loss": 0.03794408589601517
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959496717003502,
      "compression_loss": 0.0,
      "distillation_loss": 0.11199971288442612,
      "epoch": 5.25,
      "learning_rate": 7.324214169813062e-06,
      "loss": 0.108,
      "step": 5532,
      "task_loss": 0.07245731353759766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959538703817363,
      "compression_loss": 0.0,
      "distillation_loss": 0.026987843215465546,
      "epoch": 5.25,
      "learning_rate": 7.316680576204296e-06,
      "loss": 0.0251,
      "step": 5533,
      "task_loss": 0.00792054831981659
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959580661604737,
      "compression_loss": 0.0,
      "distillation_loss": 0.049572162330150604,
      "epoch": 5.26,
      "learning_rate": 7.309150194855668e-06,
      "loss": 0.0507,
      "step": 5534,
      "task_loss": 0.06096307933330536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959622590375662,
      "compression_loss": 0.0,
      "distillation_loss": 0.022030137479305267,
      "epoch": 5.26,
      "learning_rate": 7.301623027135099e-06,
      "loss": 0.0206,
      "step": 5535,
      "task_loss": 0.008088313043117523
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959664490140175,
      "compression_loss": 0.0,
      "distillation_loss": 0.2133478969335556,
      "epoch": 5.26,
      "learning_rate": 7.294099074409944e-06,
      "loss": 0.2018,
      "step": 5536,
      "task_loss": 0.09802666306495667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959706360908312,
      "compression_loss": 0.0,
      "distillation_loss": 0.020365485921502113,
      "epoch": 5.26,
      "learning_rate": 7.28657833804697e-06,
      "loss": 0.0194,
      "step": 5537,
      "task_loss": 0.01060546562075615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795974820269011,
      "compression_loss": 0.0,
      "distillation_loss": 0.04007776081562042,
      "epoch": 5.26,
      "learning_rate": 7.279060819412351e-06,
      "loss": 0.0394,
      "step": 5538,
      "task_loss": 0.03316129371523857
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959790015495606,
      "compression_loss": 0.0,
      "distillation_loss": 0.03753858804702759,
      "epoch": 5.26,
      "learning_rate": 7.271546519871672e-06,
      "loss": 0.0461,
      "step": 5539,
      "task_loss": 0.12282435595989227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959831799334837,
      "compression_loss": 0.0,
      "distillation_loss": 0.029383216053247452,
      "epoch": 5.26,
      "learning_rate": 7.264035440789954e-06,
      "loss": 0.0275,
      "step": 5540,
      "task_loss": 0.010303637012839317
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959873554217839,
      "compression_loss": 0.0,
      "distillation_loss": 0.06451848894357681,
      "epoch": 5.26,
      "learning_rate": 7.25652758353162e-06,
      "loss": 0.0609,
      "step": 5541,
      "task_loss": 0.02830340340733528
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.795991528015465,
      "compression_loss": 0.0,
      "distillation_loss": 0.07822129875421524,
      "epoch": 5.26,
      "learning_rate": 7.249022949460493e-06,
      "loss": 0.0823,
      "step": 5542,
      "task_loss": 0.1194855198264122
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959956977155306,
      "compression_loss": 0.0,
      "distillation_loss": 0.11556608974933624,
      "epoch": 5.26,
      "learning_rate": 7.2415215399398435e-06,
      "loss": 0.1179,
      "step": 5543,
      "task_loss": 0.13891972601413727
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7959998645229844,
      "compression_loss": 0.0,
      "distillation_loss": 0.017915979027748108,
      "epoch": 5.26,
      "learning_rate": 7.2340233563323284e-06,
      "loss": 0.0175,
      "step": 5544,
      "task_loss": 0.014106318354606628
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960040284388301,
      "compression_loss": 0.0,
      "distillation_loss": 0.05048777535557747,
      "epoch": 5.27,
      "learning_rate": 7.226528400000038e-06,
      "loss": 0.0469,
      "step": 5545,
      "task_loss": 0.014579741284251213
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960081894640715,
      "compression_loss": 0.0,
      "distillation_loss": 0.02436014637351036,
      "epoch": 5.27,
      "learning_rate": 7.219036672304452e-06,
      "loss": 0.0269,
      "step": 5546,
      "task_loss": 0.04929090291261673
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796012347599712,
      "compression_loss": 0.0,
      "distillation_loss": 0.1192215234041214,
      "epoch": 5.27,
      "learning_rate": 7.2115481746065e-06,
      "loss": 0.1251,
      "step": 5547,
      "task_loss": 0.17761507630348206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960165028467554,
      "compression_loss": 0.0,
      "distillation_loss": 0.05981665849685669,
      "epoch": 5.27,
      "learning_rate": 7.20406290826649e-06,
      "loss": 0.0593,
      "step": 5548,
      "task_loss": 0.054768696427345276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960206552062056,
      "compression_loss": 0.0,
      "distillation_loss": 0.07405254989862442,
      "epoch": 5.27,
      "learning_rate": 7.196580874644151e-06,
      "loss": 0.079,
      "step": 5549,
      "task_loss": 0.12388560175895691
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960248046790659,
      "compression_loss": 0.0,
      "distillation_loss": 0.0403749905526638,
      "epoch": 5.27,
      "learning_rate": 7.1891020750986475e-06,
      "loss": 0.0485,
      "step": 5550,
      "task_loss": 0.12145104259252548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960289512663403,
      "compression_loss": 0.0,
      "distillation_loss": 0.02248249761760235,
      "epoch": 5.27,
      "learning_rate": 7.181626510988529e-06,
      "loss": 0.0207,
      "step": 5551,
      "task_loss": 0.004492869600653648
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960330949690324,
      "compression_loss": 0.0,
      "distillation_loss": 0.07457023113965988,
      "epoch": 5.27,
      "learning_rate": 7.174154183671763e-06,
      "loss": 0.082,
      "step": 5552,
      "task_loss": 0.148670956492424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960372357881458,
      "compression_loss": 0.0,
      "distillation_loss": 0.06673236936330795,
      "epoch": 5.27,
      "learning_rate": 7.166685094505737e-06,
      "loss": 0.075,
      "step": 5553,
      "task_loss": 0.14950726926326752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960413737246842,
      "compression_loss": 0.0,
      "distillation_loss": 0.036908023059368134,
      "epoch": 5.27,
      "learning_rate": 7.15921924484726e-06,
      "loss": 0.052,
      "step": 5554,
      "task_loss": 0.18809108436107635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960455087796513,
      "compression_loss": 0.0,
      "distillation_loss": 0.021199818700551987,
      "epoch": 5.28,
      "learning_rate": 7.1517566360525284e-06,
      "loss": 0.0293,
      "step": 5555,
      "task_loss": 0.10172471404075623
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960496409540508,
      "compression_loss": 0.0,
      "distillation_loss": 0.10079994797706604,
      "epoch": 5.28,
      "learning_rate": 7.1442972694771545e-06,
      "loss": 0.0973,
      "step": 5556,
      "task_loss": 0.06552006304264069
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960537702488863,
      "compression_loss": 0.0,
      "distillation_loss": 0.0728619322180748,
      "epoch": 5.28,
      "learning_rate": 7.136841146476181e-06,
      "loss": 0.0749,
      "step": 5557,
      "task_loss": 0.09358760714530945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960578966651618,
      "compression_loss": 0.0,
      "distillation_loss": 0.026422590017318726,
      "epoch": 5.28,
      "learning_rate": 7.129388268404047e-06,
      "loss": 0.0249,
      "step": 5558,
      "task_loss": 0.011311305686831474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960620202038805,
      "compression_loss": 0.0,
      "distillation_loss": 0.04695506393909454,
      "epoch": 5.28,
      "learning_rate": 7.121938636614589e-06,
      "loss": 0.0484,
      "step": 5559,
      "task_loss": 0.06093965470790863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960661408660464,
      "compression_loss": 0.0,
      "distillation_loss": 0.02747797593474388,
      "epoch": 5.28,
      "learning_rate": 7.114492252461089e-06,
      "loss": 0.0339,
      "step": 5560,
      "task_loss": 0.09208115935325623
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960702586526632,
      "compression_loss": 0.0,
      "distillation_loss": 0.03135443478822708,
      "epoch": 5.28,
      "learning_rate": 7.1070491172962e-06,
      "loss": 0.0301,
      "step": 5561,
      "task_loss": 0.01891678385436535
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960743735647344,
      "compression_loss": 0.0,
      "distillation_loss": 0.060153115540742874,
      "epoch": 5.28,
      "learning_rate": 7.099609232472021e-06,
      "loss": 0.0603,
      "step": 5562,
      "task_loss": 0.06156299635767937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960784856032638,
      "compression_loss": 0.0,
      "distillation_loss": 0.02115842141211033,
      "epoch": 5.28,
      "learning_rate": 7.092172599340024e-06,
      "loss": 0.0236,
      "step": 5563,
      "task_loss": 0.045783985406160355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960825947692549,
      "compression_loss": 0.0,
      "distillation_loss": 0.016983792185783386,
      "epoch": 5.28,
      "learning_rate": 7.084739219251129e-06,
      "loss": 0.0159,
      "step": 5564,
      "task_loss": 0.005871200934052467
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960867010637117,
      "compression_loss": 0.0,
      "distillation_loss": 0.03363943099975586,
      "epoch": 5.28,
      "learning_rate": 7.0773090935556365e-06,
      "loss": 0.0311,
      "step": 5565,
      "task_loss": 0.00811656005680561
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960908044876377,
      "compression_loss": 0.0,
      "distillation_loss": 0.019596152007579803,
      "epoch": 5.29,
      "learning_rate": 7.0698822236032554e-06,
      "loss": 0.018,
      "step": 5566,
      "task_loss": 0.0034526288509368896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7960949050420365,
      "compression_loss": 0.0,
      "distillation_loss": 0.04396222531795502,
      "epoch": 5.29,
      "learning_rate": 7.0624586107431276e-06,
      "loss": 0.049,
      "step": 5567,
      "task_loss": 0.09396674484014511
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796099002727912,
      "compression_loss": 0.0,
      "distillation_loss": 0.04795515164732933,
      "epoch": 5.29,
      "learning_rate": 7.055038256323782e-06,
      "loss": 0.0508,
      "step": 5568,
      "task_loss": 0.07625043392181396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961030975462677,
      "compression_loss": 0.0,
      "distillation_loss": 0.03373149782419205,
      "epoch": 5.29,
      "learning_rate": 7.047621161693152e-06,
      "loss": 0.0355,
      "step": 5569,
      "task_loss": 0.051260001957416534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961071894981073,
      "compression_loss": 0.0,
      "distillation_loss": 0.042289115488529205,
      "epoch": 5.29,
      "learning_rate": 7.040207328198601e-06,
      "loss": 0.0427,
      "step": 5570,
      "task_loss": 0.04608525335788727
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961112785844346,
      "compression_loss": 0.0,
      "distillation_loss": 0.0194284338504076,
      "epoch": 5.29,
      "learning_rate": 7.032796757186888e-06,
      "loss": 0.0219,
      "step": 5571,
      "task_loss": 0.04372316598892212
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961153648062532,
      "compression_loss": 0.0,
      "distillation_loss": 0.028270315378904343,
      "epoch": 5.29,
      "learning_rate": 7.025389450004177e-06,
      "loss": 0.0273,
      "step": 5572,
      "task_loss": 0.018285321071743965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961194481645667,
      "compression_loss": 0.0,
      "distillation_loss": 0.11180476099252701,
      "epoch": 5.29,
      "learning_rate": 7.017985407996031e-06,
      "loss": 0.1099,
      "step": 5573,
      "task_loss": 0.09264665842056274
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961235286603789,
      "compression_loss": 0.0,
      "distillation_loss": 0.08373094350099564,
      "epoch": 5.29,
      "learning_rate": 7.010584632507444e-06,
      "loss": 0.0796,
      "step": 5574,
      "task_loss": 0.04215511679649353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961276062946936,
      "compression_loss": 0.0,
      "distillation_loss": 0.08096636831760406,
      "epoch": 5.29,
      "learning_rate": 7.0031871248827985e-06,
      "loss": 0.0826,
      "step": 5575,
      "task_loss": 0.0974457859992981
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961316810685142,
      "compression_loss": 0.0,
      "distillation_loss": 0.12849192321300507,
      "epoch": 5.3,
      "learning_rate": 6.99579288646588e-06,
      "loss": 0.1266,
      "step": 5576,
      "task_loss": 0.10947559773921967
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961357529828446,
      "compression_loss": 0.0,
      "distillation_loss": 0.08771924674510956,
      "epoch": 5.3,
      "learning_rate": 6.988401918599896e-06,
      "loss": 0.1008,
      "step": 5577,
      "task_loss": 0.2185194492340088
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961398220386883,
      "compression_loss": 0.0,
      "distillation_loss": 0.10648892819881439,
      "epoch": 5.3,
      "learning_rate": 6.981014222627444e-06,
      "loss": 0.0988,
      "step": 5578,
      "task_loss": 0.029211118817329407
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961438882370491,
      "compression_loss": 0.0,
      "distillation_loss": 0.020999478176236153,
      "epoch": 5.3,
      "learning_rate": 6.973629799890544e-06,
      "loss": 0.0194,
      "step": 5579,
      "task_loss": 0.005250850692391396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961479515789307,
      "compression_loss": 0.0,
      "distillation_loss": 0.04771939665079117,
      "epoch": 5.3,
      "learning_rate": 6.9662486517306005e-06,
      "loss": 0.0527,
      "step": 5580,
      "task_loss": 0.0980047881603241
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961520120653367,
      "compression_loss": 0.0,
      "distillation_loss": 0.051039569079875946,
      "epoch": 5.3,
      "learning_rate": 6.958870779488447e-06,
      "loss": 0.0487,
      "step": 5581,
      "task_loss": 0.027397800236940384
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961560696972708,
      "compression_loss": 0.0,
      "distillation_loss": 0.027529723942279816,
      "epoch": 5.3,
      "learning_rate": 6.951496184504306e-06,
      "loss": 0.028,
      "step": 5582,
      "task_loss": 0.03213750571012497
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961601244757368,
      "compression_loss": 0.0,
      "distillation_loss": 0.09078949689865112,
      "epoch": 5.3,
      "learning_rate": 6.944124868117796e-06,
      "loss": 0.0868,
      "step": 5583,
      "task_loss": 0.05128881335258484
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961641764017382,
      "compression_loss": 0.0,
      "distillation_loss": 0.021352669224143028,
      "epoch": 5.3,
      "learning_rate": 6.93675683166797e-06,
      "loss": 0.0196,
      "step": 5584,
      "task_loss": 0.00414588488638401
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961682254762789,
      "compression_loss": 0.0,
      "distillation_loss": 0.033476557582616806,
      "epoch": 5.3,
      "learning_rate": 6.92939207649326e-06,
      "loss": 0.0385,
      "step": 5585,
      "task_loss": 0.08368590474128723
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961722717003623,
      "compression_loss": 0.0,
      "distillation_loss": 0.025415003299713135,
      "epoch": 5.3,
      "learning_rate": 6.922030603931506e-06,
      "loss": 0.0263,
      "step": 5586,
      "task_loss": 0.03441638872027397
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961763150749924,
      "compression_loss": 0.0,
      "distillation_loss": 0.03962051123380661,
      "epoch": 5.31,
      "learning_rate": 6.914672415319945e-06,
      "loss": 0.0514,
      "step": 5587,
      "task_loss": 0.1574546843767166
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961803556011726,
      "compression_loss": 0.0,
      "distillation_loss": 0.029824821278452873,
      "epoch": 5.31,
      "learning_rate": 6.907317511995251e-06,
      "loss": 0.0368,
      "step": 5588,
      "task_loss": 0.09920459985733032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961843932799068,
      "compression_loss": 0.0,
      "distillation_loss": 0.041083455085754395,
      "epoch": 5.31,
      "learning_rate": 6.8999658952934695e-06,
      "loss": 0.0547,
      "step": 5589,
      "task_loss": 0.1776200234889984
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961884281121985,
      "compression_loss": 0.0,
      "distillation_loss": 0.07152897119522095,
      "epoch": 5.31,
      "learning_rate": 6.892617566550044e-06,
      "loss": 0.0727,
      "step": 5590,
      "task_loss": 0.0837155282497406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961924600990515,
      "compression_loss": 0.0,
      "distillation_loss": 0.015575017780065536,
      "epoch": 5.31,
      "learning_rate": 6.885272527099853e-06,
      "loss": 0.0203,
      "step": 5591,
      "task_loss": 0.06283999979496002
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7961964892414695,
      "compression_loss": 0.0,
      "distillation_loss": 0.22274234890937805,
      "epoch": 5.31,
      "learning_rate": 6.87793077827715e-06,
      "loss": 0.2181,
      "step": 5592,
      "task_loss": 0.17664700746536255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796200515540456,
      "compression_loss": 0.0,
      "distillation_loss": 0.09952651709318161,
      "epoch": 5.31,
      "learning_rate": 6.8705923214155945e-06,
      "loss": 0.1033,
      "step": 5593,
      "task_loss": 0.13760796189308167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962045389970149,
      "compression_loss": 0.0,
      "distillation_loss": 0.013694360852241516,
      "epoch": 5.31,
      "learning_rate": 6.863257157848252e-06,
      "loss": 0.0129,
      "step": 5594,
      "task_loss": 0.006061408668756485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962085596121498,
      "compression_loss": 0.0,
      "distillation_loss": 0.03154522925615311,
      "epoch": 5.31,
      "learning_rate": 6.8559252889076e-06,
      "loss": 0.0357,
      "step": 5595,
      "task_loss": 0.0730682909488678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962125773868644,
      "compression_loss": 0.0,
      "distillation_loss": 0.05232825130224228,
      "epoch": 5.31,
      "learning_rate": 6.848596715925493e-06,
      "loss": 0.0489,
      "step": 5596,
      "task_loss": 0.018482720479369164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962165923221622,
      "compression_loss": 0.0,
      "distillation_loss": 0.020783985033631325,
      "epoch": 5.32,
      "learning_rate": 6.8412714402332125e-06,
      "loss": 0.0192,
      "step": 5597,
      "task_loss": 0.0049543604254722595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962206044190472,
      "compression_loss": 0.0,
      "distillation_loss": 0.01990017667412758,
      "epoch": 5.32,
      "learning_rate": 6.833949463161438e-06,
      "loss": 0.0335,
      "step": 5598,
      "task_loss": 0.1554040163755417
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962246136785229,
      "compression_loss": 0.0,
      "distillation_loss": 0.027097290381789207,
      "epoch": 5.32,
      "learning_rate": 6.826630786040228e-06,
      "loss": 0.0248,
      "step": 5599,
      "task_loss": 0.0043129827827215195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796228620101593,
      "compression_loss": 0.0,
      "distillation_loss": 0.04166903346776962,
      "epoch": 5.32,
      "learning_rate": 6.819315410199062e-06,
      "loss": 0.0441,
      "step": 5600,
      "task_loss": 0.06640726327896118
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962326236892612,
      "compression_loss": 0.0,
      "distillation_loss": 0.02872236631810665,
      "epoch": 5.32,
      "learning_rate": 6.812003336966802e-06,
      "loss": 0.0325,
      "step": 5601,
      "task_loss": 0.06612066179513931
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962366244425311,
      "compression_loss": 0.0,
      "distillation_loss": 0.04668932408094406,
      "epoch": 5.32,
      "learning_rate": 6.8046945676717375e-06,
      "loss": 0.0442,
      "step": 5602,
      "task_loss": 0.02129777893424034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962406223624064,
      "compression_loss": 0.0,
      "distillation_loss": 0.010439068078994751,
      "epoch": 5.32,
      "learning_rate": 6.7973891036415354e-06,
      "loss": 0.0101,
      "step": 5603,
      "task_loss": 0.007036501541733742
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796244617449891,
      "compression_loss": 0.0,
      "distillation_loss": 0.017626464366912842,
      "epoch": 5.32,
      "learning_rate": 6.790086946203253e-06,
      "loss": 0.0233,
      "step": 5604,
      "task_loss": 0.07470369338989258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962486097059883,
      "compression_loss": 0.0,
      "distillation_loss": 0.1538507342338562,
      "epoch": 5.32,
      "learning_rate": 6.78278809668339e-06,
      "loss": 0.152,
      "step": 5605,
      "task_loss": 0.13536766171455383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962525991317022,
      "compression_loss": 0.0,
      "distillation_loss": 0.05390554666519165,
      "epoch": 5.32,
      "learning_rate": 6.775492556407806e-06,
      "loss": 0.054,
      "step": 5606,
      "task_loss": 0.05467413738369942
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962565857280361,
      "compression_loss": 0.0,
      "distillation_loss": 0.06552339345216751,
      "epoch": 5.32,
      "learning_rate": 6.768200326701768e-06,
      "loss": 0.0617,
      "step": 5607,
      "task_loss": 0.027765335515141487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962605694959941,
      "compression_loss": 0.0,
      "distillation_loss": 0.04036558046936989,
      "epoch": 5.33,
      "learning_rate": 6.760911408889939e-06,
      "loss": 0.0452,
      "step": 5608,
      "task_loss": 0.08895239233970642
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962645504365795,
      "compression_loss": 0.0,
      "distillation_loss": 0.14514732360839844,
      "epoch": 5.33,
      "learning_rate": 6.753625804296401e-06,
      "loss": 0.1502,
      "step": 5609,
      "task_loss": 0.19537782669067383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962685285507961,
      "compression_loss": 0.0,
      "distillation_loss": 0.04015417769551277,
      "epoch": 5.33,
      "learning_rate": 6.746343514244611e-06,
      "loss": 0.0367,
      "step": 5610,
      "task_loss": 0.006075270473957062
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962725038396478,
      "compression_loss": 0.0,
      "distillation_loss": 0.07076828181743622,
      "epoch": 5.33,
      "learning_rate": 6.739064540057424e-06,
      "loss": 0.0669,
      "step": 5611,
      "task_loss": 0.03183208778500557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962764763041379,
      "compression_loss": 0.0,
      "distillation_loss": 0.0222453735768795,
      "epoch": 5.33,
      "learning_rate": 6.731788883057116e-06,
      "loss": 0.0287,
      "step": 5612,
      "task_loss": 0.08726339787244797
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962804459452703,
      "compression_loss": 0.0,
      "distillation_loss": 0.03244253247976303,
      "epoch": 5.33,
      "learning_rate": 6.724516544565332e-06,
      "loss": 0.0367,
      "step": 5613,
      "task_loss": 0.07509634643793106
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962844127640487,
      "compression_loss": 0.0,
      "distillation_loss": 0.017709966748952866,
      "epoch": 5.33,
      "learning_rate": 6.717247525903142e-06,
      "loss": 0.0164,
      "step": 5614,
      "task_loss": 0.004615955054759979
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962883767614767,
      "compression_loss": 0.0,
      "distillation_loss": 0.04239179193973541,
      "epoch": 5.33,
      "learning_rate": 6.709981828390979e-06,
      "loss": 0.0433,
      "step": 5615,
      "task_loss": 0.051517605781555176
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796292337938558,
      "compression_loss": 0.0,
      "distillation_loss": 0.0392778217792511,
      "epoch": 5.33,
      "learning_rate": 6.70271945334871e-06,
      "loss": 0.0433,
      "step": 5616,
      "task_loss": 0.07987174391746521
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7962962962962964,
      "compression_loss": 0.0,
      "distillation_loss": 0.05544783174991608,
      "epoch": 5.33,
      "learning_rate": 6.695460402095577e-06,
      "loss": 0.0553,
      "step": 5617,
      "task_loss": 0.0541701577603817
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963002518356953,
      "compression_loss": 0.0,
      "distillation_loss": 0.025989163666963577,
      "epoch": 5.34,
      "learning_rate": 6.688204675950205e-06,
      "loss": 0.0389,
      "step": 5618,
      "task_loss": 0.15557032823562622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963042045577587,
      "compression_loss": 0.0,
      "distillation_loss": 0.029921121895313263,
      "epoch": 5.34,
      "learning_rate": 6.6809522762306566e-06,
      "loss": 0.0329,
      "step": 5619,
      "task_loss": 0.05976049602031708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963081544634901,
      "compression_loss": 0.0,
      "distillation_loss": 0.044836804270744324,
      "epoch": 5.34,
      "learning_rate": 6.673703204254347e-06,
      "loss": 0.0436,
      "step": 5620,
      "task_loss": 0.03275785595178604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963121015538933,
      "compression_loss": 0.0,
      "distillation_loss": 0.037071071565151215,
      "epoch": 5.34,
      "learning_rate": 6.666457461338108e-06,
      "loss": 0.0357,
      "step": 5621,
      "task_loss": 0.023638306185603142
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963160458299718,
      "compression_loss": 0.0,
      "distillation_loss": 0.0316130630671978,
      "epoch": 5.34,
      "learning_rate": 6.659215048798164e-06,
      "loss": 0.0289,
      "step": 5622,
      "task_loss": 0.004098406061530113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963199872927296,
      "compression_loss": 0.0,
      "distillation_loss": 0.021150682121515274,
      "epoch": 5.34,
      "learning_rate": 6.651975967950147e-06,
      "loss": 0.0196,
      "step": 5623,
      "task_loss": 0.005362769588828087
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79632392594317,
      "compression_loss": 0.0,
      "distillation_loss": 0.029035231098532677,
      "epoch": 5.34,
      "learning_rate": 6.644740220109058e-06,
      "loss": 0.0267,
      "step": 5624,
      "task_loss": 0.0061414651572704315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963278617822969,
      "compression_loss": 0.0,
      "distillation_loss": 0.06975620985031128,
      "epoch": 5.34,
      "learning_rate": 6.6375078065893e-06,
      "loss": 0.0782,
      "step": 5625,
      "task_loss": 0.15371599793434143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963317948111139,
      "compression_loss": 0.0,
      "distillation_loss": 0.06846155226230621,
      "epoch": 5.34,
      "learning_rate": 6.630278728704692e-06,
      "loss": 0.0738,
      "step": 5626,
      "task_loss": 0.12186034023761749
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963357250306248,
      "compression_loss": 0.0,
      "distillation_loss": 0.05929230898618698,
      "epoch": 5.34,
      "learning_rate": 6.6230529877684215e-06,
      "loss": 0.0548,
      "step": 5627,
      "task_loss": 0.014448923990130424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963396524418331,
      "compression_loss": 0.0,
      "distillation_loss": 0.032660387456417084,
      "epoch": 5.34,
      "learning_rate": 6.615830585093074e-06,
      "loss": 0.0351,
      "step": 5628,
      "task_loss": 0.05667547136545181
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963435770457427,
      "compression_loss": 0.0,
      "distillation_loss": 0.01413150317966938,
      "epoch": 5.35,
      "learning_rate": 6.6086115219906485e-06,
      "loss": 0.0132,
      "step": 5629,
      "task_loss": 0.004492944106459618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963474988433571,
      "compression_loss": 0.0,
      "distillation_loss": 0.11651385575532913,
      "epoch": 5.35,
      "learning_rate": 6.601395799772503e-06,
      "loss": 0.1489,
      "step": 5630,
      "task_loss": 0.4406876266002655
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79635141783568,
      "compression_loss": 0.0,
      "distillation_loss": 0.08212954550981522,
      "epoch": 5.35,
      "learning_rate": 6.594183419749431e-06,
      "loss": 0.0788,
      "step": 5631,
      "task_loss": 0.048624441027641296
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963553340237152,
      "compression_loss": 0.0,
      "distillation_loss": 0.024417495355010033,
      "epoch": 5.35,
      "learning_rate": 6.586974383231573e-06,
      "loss": 0.0427,
      "step": 5632,
      "task_loss": 0.20715758204460144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963592474084663,
      "compression_loss": 0.0,
      "distillation_loss": 0.07218081504106522,
      "epoch": 5.35,
      "learning_rate": 6.579768691528504e-06,
      "loss": 0.0702,
      "step": 5633,
      "task_loss": 0.052654679864645004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796363157990937,
      "compression_loss": 0.0,
      "distillation_loss": 0.04733956977725029,
      "epoch": 5.35,
      "learning_rate": 6.572566345949166e-06,
      "loss": 0.0439,
      "step": 5634,
      "task_loss": 0.012745276093482971
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796367065772131,
      "compression_loss": 0.0,
      "distillation_loss": 0.02717157080769539,
      "epoch": 5.35,
      "learning_rate": 6.565367347801893e-06,
      "loss": 0.0268,
      "step": 5635,
      "task_loss": 0.023610301315784454
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796370970753052,
      "compression_loss": 0.0,
      "distillation_loss": 0.029316646978259087,
      "epoch": 5.35,
      "learning_rate": 6.5581716983944274e-06,
      "loss": 0.0272,
      "step": 5636,
      "task_loss": 0.008220043033361435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963748729347035,
      "compression_loss": 0.0,
      "distillation_loss": 0.12085194140672684,
      "epoch": 5.35,
      "learning_rate": 6.550979399033894e-06,
      "loss": 0.1194,
      "step": 5637,
      "task_loss": 0.10640271008014679
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963787723180894,
      "compression_loss": 0.0,
      "distillation_loss": 0.033697567880153656,
      "epoch": 5.35,
      "learning_rate": 6.5437904510267935e-06,
      "loss": 0.0313,
      "step": 5638,
      "task_loss": 0.010095924139022827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963826689042134,
      "compression_loss": 0.0,
      "distillation_loss": 0.14860667288303375,
      "epoch": 5.36,
      "learning_rate": 6.536604855679043e-06,
      "loss": 0.1594,
      "step": 5639,
      "task_loss": 0.2568969428539276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963865626940789,
      "compression_loss": 0.0,
      "distillation_loss": 0.030187159776687622,
      "epoch": 5.36,
      "learning_rate": 6.529422614295949e-06,
      "loss": 0.0275,
      "step": 5640,
      "task_loss": 0.003732619807124138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79639045368869,
      "compression_loss": 0.0,
      "distillation_loss": 0.024757564067840576,
      "epoch": 5.36,
      "learning_rate": 6.522243728182195e-06,
      "loss": 0.0257,
      "step": 5641,
      "task_loss": 0.03456170856952667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79639434188905,
      "compression_loss": 0.0,
      "distillation_loss": 0.07199036329984665,
      "epoch": 5.36,
      "learning_rate": 6.5150681986418466e-06,
      "loss": 0.0716,
      "step": 5642,
      "task_loss": 0.06780923902988434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7963982272961627,
      "compression_loss": 0.0,
      "distillation_loss": 0.02806227095425129,
      "epoch": 5.36,
      "learning_rate": 6.507896026978394e-06,
      "loss": 0.0354,
      "step": 5643,
      "task_loss": 0.1011197566986084
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964021099110319,
      "compression_loss": 0.0,
      "distillation_loss": 0.05786515027284622,
      "epoch": 5.36,
      "learning_rate": 6.500727214494687e-06,
      "loss": 0.0632,
      "step": 5644,
      "task_loss": 0.11146190017461777
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964059897346611,
      "compression_loss": 0.0,
      "distillation_loss": 0.05632102116942406,
      "epoch": 5.36,
      "learning_rate": 6.493561762492966e-06,
      "loss": 0.077,
      "step": 5645,
      "task_loss": 0.2627117931842804
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964098667680543,
      "compression_loss": 0.0,
      "distillation_loss": 0.05039035528898239,
      "epoch": 5.36,
      "learning_rate": 6.48639967227489e-06,
      "loss": 0.056,
      "step": 5646,
      "task_loss": 0.10663844645023346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964137410122147,
      "compression_loss": 0.0,
      "distillation_loss": 0.057316750288009644,
      "epoch": 5.36,
      "learning_rate": 6.4792409451414735e-06,
      "loss": 0.0602,
      "step": 5647,
      "task_loss": 0.08620242029428482
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964176124681465,
      "compression_loss": 0.0,
      "distillation_loss": 0.04628031328320503,
      "epoch": 5.36,
      "learning_rate": 6.472085582393128e-06,
      "loss": 0.0488,
      "step": 5648,
      "task_loss": 0.07143149524927139
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796421481136853,
      "compression_loss": 0.0,
      "distillation_loss": 0.1507255733013153,
      "epoch": 5.36,
      "learning_rate": 6.4649335853296685e-06,
      "loss": 0.1469,
      "step": 5649,
      "task_loss": 0.11250782012939453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796425347019338,
      "compression_loss": 0.0,
      "distillation_loss": 0.07052607089281082,
      "epoch": 5.37,
      "learning_rate": 6.457784955250296e-06,
      "loss": 0.077,
      "step": 5650,
      "task_loss": 0.13528487086296082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964292101166052,
      "compression_loss": 0.0,
      "distillation_loss": 0.012097038328647614,
      "epoch": 5.37,
      "learning_rate": 6.450639693453589e-06,
      "loss": 0.0113,
      "step": 5651,
      "task_loss": 0.004289904609322548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964330704296583,
      "compression_loss": 0.0,
      "distillation_loss": 0.03070550039410591,
      "epoch": 5.37,
      "learning_rate": 6.443497801237505e-06,
      "loss": 0.0397,
      "step": 5652,
      "task_loss": 0.12050429731607437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964369279595008,
      "compression_loss": 0.0,
      "distillation_loss": 0.02167482301592827,
      "epoch": 5.37,
      "learning_rate": 6.436359279899426e-06,
      "loss": 0.0288,
      "step": 5653,
      "task_loss": 0.09340011328458786
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964407827071367,
      "compression_loss": 0.0,
      "distillation_loss": 0.03893493860960007,
      "epoch": 5.37,
      "learning_rate": 6.429224130736084e-06,
      "loss": 0.0371,
      "step": 5654,
      "task_loss": 0.020514240488409996
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964446346735695,
      "compression_loss": 0.0,
      "distillation_loss": 0.06006263196468353,
      "epoch": 5.37,
      "learning_rate": 6.4220923550436106e-06,
      "loss": 0.0578,
      "step": 5655,
      "task_loss": 0.0373433418571949
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964484838598029,
      "compression_loss": 0.0,
      "distillation_loss": 0.05972398817539215,
      "epoch": 5.37,
      "learning_rate": 6.414963954117534e-06,
      "loss": 0.0573,
      "step": 5656,
      "task_loss": 0.035765524953603745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964523302668406,
      "compression_loss": 0.0,
      "distillation_loss": 0.11887294799089432,
      "epoch": 5.37,
      "learning_rate": 6.407838929252768e-06,
      "loss": 0.114,
      "step": 5657,
      "task_loss": 0.07028041779994965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964561738956862,
      "compression_loss": 0.0,
      "distillation_loss": 0.03886423259973526,
      "epoch": 5.37,
      "learning_rate": 6.400717281743601e-06,
      "loss": 0.0367,
      "step": 5658,
      "task_loss": 0.017281973734498024
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964600147473435,
      "compression_loss": 0.0,
      "distillation_loss": 0.023825645446777344,
      "epoch": 5.37,
      "learning_rate": 6.393599012883708e-06,
      "loss": 0.0297,
      "step": 5659,
      "task_loss": 0.08249877393245697
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964638528228162,
      "compression_loss": 0.0,
      "distillation_loss": 0.029711570590734482,
      "epoch": 5.38,
      "learning_rate": 6.386484123966171e-06,
      "loss": 0.0276,
      "step": 5660,
      "task_loss": 0.00878220796585083
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964676881231078,
      "compression_loss": 0.0,
      "distillation_loss": 0.09236393123865128,
      "epoch": 5.38,
      "learning_rate": 6.379372616283436e-06,
      "loss": 0.1032,
      "step": 5661,
      "task_loss": 0.20039357244968414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964715206492221,
      "compression_loss": 0.0,
      "distillation_loss": 0.045773375779390335,
      "epoch": 5.38,
      "learning_rate": 6.372264491127336e-06,
      "loss": 0.0429,
      "step": 5662,
      "task_loss": 0.01664689928293228
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964753504021628,
      "compression_loss": 0.0,
      "distillation_loss": 0.058652136474847794,
      "epoch": 5.38,
      "learning_rate": 6.365159749789112e-06,
      "loss": 0.0581,
      "step": 5663,
      "task_loss": 0.05268477275967598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964791773829336,
      "compression_loss": 0.0,
      "distillation_loss": 0.07769237458705902,
      "epoch": 5.38,
      "learning_rate": 6.358058393559366e-06,
      "loss": 0.0717,
      "step": 5664,
      "task_loss": 0.017609622329473495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796483001592538,
      "compression_loss": 0.0,
      "distillation_loss": 0.01842435821890831,
      "epoch": 5.38,
      "learning_rate": 6.350960423728083e-06,
      "loss": 0.024,
      "step": 5665,
      "task_loss": 0.07376164942979813
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79648682303198,
      "compression_loss": 0.0,
      "distillation_loss": 0.03854244574904442,
      "epoch": 5.38,
      "learning_rate": 6.3438658415846565e-06,
      "loss": 0.0351,
      "step": 5666,
      "task_loss": 0.0036998502910137177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796490641702263,
      "compression_loss": 0.0,
      "distillation_loss": 0.018844418227672577,
      "epoch": 5.38,
      "learning_rate": 6.336774648417854e-06,
      "loss": 0.0237,
      "step": 5667,
      "task_loss": 0.06747782230377197
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964944576043909,
      "compression_loss": 0.0,
      "distillation_loss": 0.03793220967054367,
      "epoch": 5.38,
      "learning_rate": 6.329686845515823e-06,
      "loss": 0.0457,
      "step": 5668,
      "task_loss": 0.11540687829256058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7964982707393672,
      "compression_loss": 0.0,
      "distillation_loss": 0.06492698937654495,
      "epoch": 5.38,
      "learning_rate": 6.322602434166083e-06,
      "loss": 0.0729,
      "step": 5669,
      "task_loss": 0.14457391202449799
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965020811081956,
      "compression_loss": 0.0,
      "distillation_loss": 0.08671444654464722,
      "epoch": 5.38,
      "learning_rate": 6.315521415655571e-06,
      "loss": 0.0834,
      "step": 5670,
      "task_loss": 0.05390065908432007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965058887118799,
      "compression_loss": 0.0,
      "distillation_loss": 0.03182661160826683,
      "epoch": 5.39,
      "learning_rate": 6.308443791270579e-06,
      "loss": 0.0292,
      "step": 5671,
      "task_loss": 0.005619386211037636
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965096935514236,
      "compression_loss": 0.0,
      "distillation_loss": 0.019694361835718155,
      "epoch": 5.39,
      "learning_rate": 6.301369562296786e-06,
      "loss": 0.0201,
      "step": 5672,
      "task_loss": 0.02362089231610298
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965134956278307,
      "compression_loss": 0.0,
      "distillation_loss": 0.09567753225564957,
      "epoch": 5.39,
      "learning_rate": 6.294298730019271e-06,
      "loss": 0.1003,
      "step": 5673,
      "task_loss": 0.14172454178333282
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965172949421045,
      "compression_loss": 0.0,
      "distillation_loss": 0.07337604463100433,
      "epoch": 5.39,
      "learning_rate": 6.28723129572247e-06,
      "loss": 0.0814,
      "step": 5674,
      "task_loss": 0.1531829535961151
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796521091495249,
      "compression_loss": 0.0,
      "distillation_loss": 0.029669515788555145,
      "epoch": 5.39,
      "learning_rate": 6.280167260690237e-06,
      "loss": 0.0346,
      "step": 5675,
      "task_loss": 0.07867026329040527
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965248852882676,
      "compression_loss": 0.0,
      "distillation_loss": 0.02858271822333336,
      "epoch": 5.39,
      "learning_rate": 6.273106626205768e-06,
      "loss": 0.0355,
      "step": 5676,
      "task_loss": 0.09754221141338348
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965286763221643,
      "compression_loss": 0.0,
      "distillation_loss": 0.058425020426511765,
      "epoch": 5.39,
      "learning_rate": 6.266049393551679e-06,
      "loss": 0.0646,
      "step": 5677,
      "task_loss": 0.1198449432849884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965324645979425,
      "compression_loss": 0.0,
      "distillation_loss": 0.030206482857465744,
      "epoch": 5.39,
      "learning_rate": 6.258995564009939e-06,
      "loss": 0.0289,
      "step": 5678,
      "task_loss": 0.017161451280117035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796536250116606,
      "compression_loss": 0.0,
      "distillation_loss": 0.016308438032865524,
      "epoch": 5.39,
      "learning_rate": 6.251945138861915e-06,
      "loss": 0.0206,
      "step": 5679,
      "task_loss": 0.0590689480304718
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965400328791584,
      "compression_loss": 0.0,
      "distillation_loss": 0.06689819693565369,
      "epoch": 5.39,
      "learning_rate": 6.244898119388337e-06,
      "loss": 0.0614,
      "step": 5680,
      "task_loss": 0.011653106659650803
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965438128866036,
      "compression_loss": 0.0,
      "distillation_loss": 0.047350913286209106,
      "epoch": 5.4,
      "learning_rate": 6.2378545068693505e-06,
      "loss": 0.0555,
      "step": 5681,
      "task_loss": 0.1290304809808731
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796547590139945,
      "compression_loss": 0.0,
      "distillation_loss": 0.026393115520477295,
      "epoch": 5.4,
      "learning_rate": 6.230814302584445e-06,
      "loss": 0.0243,
      "step": 5682,
      "task_loss": 0.005752213299274445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965513646401865,
      "compression_loss": 0.0,
      "distillation_loss": 0.023942215368151665,
      "epoch": 5.4,
      "learning_rate": 6.223777507812514e-06,
      "loss": 0.0303,
      "step": 5683,
      "task_loss": 0.08767253160476685
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965551363883316,
      "compression_loss": 0.0,
      "distillation_loss": 0.17860738933086395,
      "epoch": 5.4,
      "learning_rate": 6.216744123831836e-06,
      "loss": 0.1826,
      "step": 5684,
      "task_loss": 0.21895524859428406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965589053853842,
      "compression_loss": 0.0,
      "distillation_loss": 0.07975465059280396,
      "epoch": 5.4,
      "learning_rate": 6.209714151920046e-06,
      "loss": 0.0936,
      "step": 5685,
      "task_loss": 0.21826988458633423
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965626716323478,
      "compression_loss": 0.0,
      "distillation_loss": 0.03170043230056763,
      "epoch": 5.4,
      "learning_rate": 6.2026875933541785e-06,
      "loss": 0.0301,
      "step": 5686,
      "task_loss": 0.015506980940699577
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965664351302263,
      "compression_loss": 0.0,
      "distillation_loss": 0.05559496954083443,
      "epoch": 5.4,
      "learning_rate": 6.195664449410629e-06,
      "loss": 0.0656,
      "step": 5687,
      "task_loss": 0.15556418895721436
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965701958800231,
      "compression_loss": 0.0,
      "distillation_loss": 0.0222992654889822,
      "epoch": 5.4,
      "learning_rate": 6.188644721365203e-06,
      "loss": 0.0213,
      "step": 5688,
      "task_loss": 0.012770412489771843
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796573953882742,
      "compression_loss": 0.0,
      "distillation_loss": 0.20726126432418823,
      "epoch": 5.4,
      "learning_rate": 6.181628410493059e-06,
      "loss": 0.1953,
      "step": 5689,
      "task_loss": 0.08764246851205826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965777091393867,
      "compression_loss": 0.0,
      "distillation_loss": 0.05251915007829666,
      "epoch": 5.4,
      "learning_rate": 6.174615518068738e-06,
      "loss": 0.0573,
      "step": 5690,
      "task_loss": 0.10022127628326416
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965814616509609,
      "compression_loss": 0.0,
      "distillation_loss": 0.021990936249494553,
      "epoch": 5.4,
      "learning_rate": 6.16760604536617e-06,
      "loss": 0.022,
      "step": 5691,
      "task_loss": 0.02201257273554802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965852114184683,
      "compression_loss": 0.0,
      "distillation_loss": 0.02736847475171089,
      "epoch": 5.41,
      "learning_rate": 6.1605999936586725e-06,
      "loss": 0.0252,
      "step": 5692,
      "task_loss": 0.005297476425766945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965889584429124,
      "compression_loss": 0.0,
      "distillation_loss": 0.02012113854289055,
      "epoch": 5.41,
      "learning_rate": 6.153597364218916e-06,
      "loss": 0.0185,
      "step": 5693,
      "task_loss": 0.004320105537772179
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7965927027252971,
      "compression_loss": 0.0,
      "distillation_loss": 0.05933302640914917,
      "epoch": 5.41,
      "learning_rate": 6.146598158318956e-06,
      "loss": 0.0612,
      "step": 5694,
      "task_loss": 0.07751144468784332
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796596444266626,
      "compression_loss": 0.0,
      "distillation_loss": 0.1531362384557724,
      "epoch": 5.41,
      "learning_rate": 6.1396023772302465e-06,
      "loss": 0.1486,
      "step": 5695,
      "task_loss": 0.10774153470993042
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966001830679028,
      "compression_loss": 0.0,
      "distillation_loss": 0.032585859298706055,
      "epoch": 5.41,
      "learning_rate": 6.132610022223598e-06,
      "loss": 0.0408,
      "step": 5696,
      "task_loss": 0.11497241258621216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966039191301312,
      "compression_loss": 0.0,
      "distillation_loss": 0.13292592763900757,
      "epoch": 5.41,
      "learning_rate": 6.125621094569198e-06,
      "loss": 0.1305,
      "step": 5697,
      "task_loss": 0.10847350209951401
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966076524543149,
      "compression_loss": 0.0,
      "distillation_loss": 0.036475587636232376,
      "epoch": 5.41,
      "learning_rate": 6.118635595536634e-06,
      "loss": 0.0353,
      "step": 5698,
      "task_loss": 0.024255897849798203
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966113830414574,
      "compression_loss": 0.0,
      "distillation_loss": 0.018289241939783096,
      "epoch": 5.41,
      "learning_rate": 6.111653526394839e-06,
      "loss": 0.0234,
      "step": 5699,
      "task_loss": 0.06973688304424286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966151108925625,
      "compression_loss": 0.0,
      "distillation_loss": 0.11107459664344788,
      "epoch": 5.41,
      "learning_rate": 6.104674888412157e-06,
      "loss": 0.114,
      "step": 5700,
      "task_loss": 0.1403195559978485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796618836008634,
      "compression_loss": 0.0,
      "distillation_loss": 0.03960804641246796,
      "epoch": 5.41,
      "learning_rate": 6.097699682856275e-06,
      "loss": 0.0449,
      "step": 5701,
      "task_loss": 0.09255840629339218
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966225583906754,
      "compression_loss": 0.0,
      "distillation_loss": 0.06171734258532524,
      "epoch": 5.42,
      "learning_rate": 6.090727910994287e-06,
      "loss": 0.0701,
      "step": 5702,
      "task_loss": 0.1456369161605835
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966262780396905,
      "compression_loss": 0.0,
      "distillation_loss": 0.02067572996020317,
      "epoch": 5.42,
      "learning_rate": 6.083759574092643e-06,
      "loss": 0.019,
      "step": 5703,
      "task_loss": 0.00421629473567009
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796629994956683,
      "compression_loss": 0.0,
      "distillation_loss": 0.024341976270079613,
      "epoch": 5.42,
      "learning_rate": 6.076794673417166e-06,
      "loss": 0.0314,
      "step": 5704,
      "task_loss": 0.09478569775819778
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966337091426565,
      "compression_loss": 0.0,
      "distillation_loss": 0.00914972648024559,
      "epoch": 5.42,
      "learning_rate": 6.069833210233078e-06,
      "loss": 0.0086,
      "step": 5705,
      "task_loss": 0.0033750738948583603
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966374205986146,
      "compression_loss": 0.0,
      "distillation_loss": 0.08902280032634735,
      "epoch": 5.42,
      "learning_rate": 6.062875185804958e-06,
      "loss": 0.0931,
      "step": 5706,
      "task_loss": 0.13023674488067627
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966411293255612,
      "compression_loss": 0.0,
      "distillation_loss": 0.01705138012766838,
      "epoch": 5.42,
      "learning_rate": 6.055920601396753e-06,
      "loss": 0.025,
      "step": 5707,
      "task_loss": 0.09632694721221924
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966448353244998,
      "compression_loss": 0.0,
      "distillation_loss": 0.03472287952899933,
      "epoch": 5.42,
      "learning_rate": 6.048969458271808e-06,
      "loss": 0.0329,
      "step": 5708,
      "task_loss": 0.01673199236392975
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966485385964343,
      "compression_loss": 0.0,
      "distillation_loss": 0.08574015647172928,
      "epoch": 5.42,
      "learning_rate": 6.0420217576928365e-06,
      "loss": 0.0943,
      "step": 5709,
      "task_loss": 0.17116937041282654
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796652239142368,
      "compression_loss": 0.0,
      "distillation_loss": 0.02010597288608551,
      "epoch": 5.42,
      "learning_rate": 6.035077500921918e-06,
      "loss": 0.0189,
      "step": 5710,
      "task_loss": 0.008190853521227837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796655936963305,
      "compression_loss": 0.0,
      "distillation_loss": 0.05432071536779404,
      "epoch": 5.42,
      "learning_rate": 6.028136689220498e-06,
      "loss": 0.0523,
      "step": 5711,
      "task_loss": 0.03373948112130165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966596320602487,
      "compression_loss": 0.0,
      "distillation_loss": 0.027371380478143692,
      "epoch": 5.42,
      "learning_rate": 6.021199323849424e-06,
      "loss": 0.032,
      "step": 5712,
      "task_loss": 0.0741061344742775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796663324434203,
      "compression_loss": 0.0,
      "distillation_loss": 0.021952230483293533,
      "epoch": 5.43,
      "learning_rate": 6.014265406068897e-06,
      "loss": 0.0308,
      "step": 5713,
      "task_loss": 0.1101214811205864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966670140861714,
      "compression_loss": 0.0,
      "distillation_loss": 0.04616188630461693,
      "epoch": 5.43,
      "learning_rate": 6.00733493713849e-06,
      "loss": 0.0514,
      "step": 5714,
      "task_loss": 0.09814758598804474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966707010171575,
      "compression_loss": 0.0,
      "distillation_loss": 0.0602332204580307,
      "epoch": 5.43,
      "learning_rate": 6.000407918317167e-06,
      "loss": 0.0716,
      "step": 5715,
      "task_loss": 0.17372044920921326
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966743852281654,
      "compression_loss": 0.0,
      "distillation_loss": 0.11245335638523102,
      "epoch": 5.43,
      "learning_rate": 5.993484350863246e-06,
      "loss": 0.1094,
      "step": 5716,
      "task_loss": 0.0819602832198143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966780667201983,
      "compression_loss": 0.0,
      "distillation_loss": 0.033800143748521805,
      "epoch": 5.43,
      "learning_rate": 5.986564236034426e-06,
      "loss": 0.0379,
      "step": 5717,
      "task_loss": 0.07496166974306107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966817454942602,
      "compression_loss": 0.0,
      "distillation_loss": 0.03140062838792801,
      "epoch": 5.43,
      "learning_rate": 5.9796475750877795e-06,
      "loss": 0.0294,
      "step": 5718,
      "task_loss": 0.011493334546685219
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966854215513546,
      "compression_loss": 0.0,
      "distillation_loss": 0.14226366579532623,
      "epoch": 5.43,
      "learning_rate": 5.9727343692797615e-06,
      "loss": 0.153,
      "step": 5719,
      "task_loss": 0.24922578036785126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966890948924853,
      "compression_loss": 0.0,
      "distillation_loss": 0.04390806332230568,
      "epoch": 5.43,
      "learning_rate": 5.965824619866184e-06,
      "loss": 0.0484,
      "step": 5720,
      "task_loss": 0.08851758390665054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796692765518656,
      "compression_loss": 0.0,
      "distillation_loss": 0.05784032121300697,
      "epoch": 5.43,
      "learning_rate": 5.958918328102223e-06,
      "loss": 0.0751,
      "step": 5721,
      "task_loss": 0.2306969314813614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7966964334308702,
      "compression_loss": 0.0,
      "distillation_loss": 0.11825370043516159,
      "epoch": 5.43,
      "learning_rate": 5.9520154952424606e-06,
      "loss": 0.1196,
      "step": 5722,
      "task_loss": 0.13163554668426514
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967000986301317,
      "compression_loss": 0.0,
      "distillation_loss": 0.03910898417234421,
      "epoch": 5.43,
      "learning_rate": 5.945116122540817e-06,
      "loss": 0.046,
      "step": 5723,
      "task_loss": 0.10799519717693329
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967037611174442,
      "compression_loss": 0.0,
      "distillation_loss": 0.03271537274122238,
      "epoch": 5.44,
      "learning_rate": 5.938220211250595e-06,
      "loss": 0.03,
      "step": 5724,
      "task_loss": 0.005857875570654869
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967074208938114,
      "compression_loss": 0.0,
      "distillation_loss": 0.017977934330701828,
      "epoch": 5.44,
      "learning_rate": 5.9313277626244725e-06,
      "loss": 0.0242,
      "step": 5725,
      "task_loss": 0.0803464874625206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796711077960237,
      "compression_loss": 0.0,
      "distillation_loss": 0.0364721417427063,
      "epoch": 5.44,
      "learning_rate": 5.924438777914504e-06,
      "loss": 0.0424,
      "step": 5726,
      "task_loss": 0.0960368812084198
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967147323177246,
      "compression_loss": 0.0,
      "distillation_loss": 0.04610713943839073,
      "epoch": 5.44,
      "learning_rate": 5.917553258372102e-06,
      "loss": 0.0538,
      "step": 5727,
      "task_loss": 0.12283903360366821
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967183839672779,
      "compression_loss": 0.0,
      "distillation_loss": 0.06002458930015564,
      "epoch": 5.44,
      "learning_rate": 5.910671205248045e-06,
      "loss": 0.0667,
      "step": 5728,
      "task_loss": 0.12689921259880066
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967220329099006,
      "compression_loss": 0.0,
      "distillation_loss": 0.11154115200042725,
      "epoch": 5.44,
      "learning_rate": 5.903792619792506e-06,
      "loss": 0.1008,
      "step": 5729,
      "task_loss": 0.004503896459937096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967256791465963,
      "compression_loss": 0.0,
      "distillation_loss": 0.03865112364292145,
      "epoch": 5.44,
      "learning_rate": 5.896917503255006e-06,
      "loss": 0.037,
      "step": 5730,
      "task_loss": 0.022370828315615654
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967293226783689,
      "compression_loss": 0.0,
      "distillation_loss": 0.02259804867208004,
      "epoch": 5.44,
      "learning_rate": 5.890045856884435e-06,
      "loss": 0.0404,
      "step": 5731,
      "task_loss": 0.20083081722259521
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967329635062219,
      "compression_loss": 0.0,
      "distillation_loss": 0.03995134308934212,
      "epoch": 5.44,
      "learning_rate": 5.883177681929078e-06,
      "loss": 0.0678,
      "step": 5732,
      "task_loss": 0.31826668977737427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796736601631159,
      "compression_loss": 0.0,
      "distillation_loss": 0.038897112011909485,
      "epoch": 5.44,
      "learning_rate": 5.876312979636561e-06,
      "loss": 0.0417,
      "step": 5733,
      "task_loss": 0.06685265898704529
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967402370541838,
      "compression_loss": 0.0,
      "distillation_loss": 0.018638338893651962,
      "epoch": 5.45,
      "learning_rate": 5.869451751253885e-06,
      "loss": 0.0171,
      "step": 5734,
      "task_loss": 0.0036374125629663467
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967438697763003,
      "compression_loss": 0.0,
      "distillation_loss": 0.03524959087371826,
      "epoch": 5.45,
      "learning_rate": 5.8625939980274295e-06,
      "loss": 0.0405,
      "step": 5735,
      "task_loss": 0.0877641811966896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967474997985118,
      "compression_loss": 0.0,
      "distillation_loss": 0.0975470244884491,
      "epoch": 5.45,
      "learning_rate": 5.855739721202952e-06,
      "loss": 0.0919,
      "step": 5736,
      "task_loss": 0.04134809970855713
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967511271218222,
      "compression_loss": 0.0,
      "distillation_loss": 0.047942765057086945,
      "epoch": 5.45,
      "learning_rate": 5.848888922025553e-06,
      "loss": 0.0523,
      "step": 5737,
      "task_loss": 0.09145848453044891
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967547517472352,
      "compression_loss": 0.0,
      "distillation_loss": 0.05359815061092377,
      "epoch": 5.45,
      "learning_rate": 5.842041601739706e-06,
      "loss": 0.0587,
      "step": 5738,
      "task_loss": 0.10473354160785675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967583736757543,
      "compression_loss": 0.0,
      "distillation_loss": 0.07147708535194397,
      "epoch": 5.45,
      "learning_rate": 5.835197761589275e-06,
      "loss": 0.0859,
      "step": 5739,
      "task_loss": 0.215956911444664
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967619929083833,
      "compression_loss": 0.0,
      "distillation_loss": 0.047319263219833374,
      "epoch": 5.45,
      "learning_rate": 5.828357402817469e-06,
      "loss": 0.0604,
      "step": 5740,
      "task_loss": 0.17839795351028442
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796765609446126,
      "compression_loss": 0.0,
      "distillation_loss": 0.12125735729932785,
      "epoch": 5.45,
      "learning_rate": 5.821520526666868e-06,
      "loss": 0.1155,
      "step": 5741,
      "task_loss": 0.06390950828790665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967692232899859,
      "compression_loss": 0.0,
      "distillation_loss": 0.02730424515902996,
      "epoch": 5.45,
      "learning_rate": 5.8146871343794315e-06,
      "loss": 0.0254,
      "step": 5742,
      "task_loss": 0.00804077833890915
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967728344409667,
      "compression_loss": 0.0,
      "distillation_loss": 0.07533392310142517,
      "epoch": 5.45,
      "learning_rate": 5.80785722719647e-06,
      "loss": 0.0823,
      "step": 5743,
      "task_loss": 0.14540576934814453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967764429000722,
      "compression_loss": 0.0,
      "distillation_loss": 0.1283230483531952,
      "epoch": 5.45,
      "learning_rate": 5.801030806358679e-06,
      "loss": 0.1246,
      "step": 5744,
      "task_loss": 0.09111975133419037
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796780048668306,
      "compression_loss": 0.0,
      "distillation_loss": 0.07009841501712799,
      "epoch": 5.46,
      "learning_rate": 5.7942078731061e-06,
      "loss": 0.0717,
      "step": 5745,
      "task_loss": 0.08628898113965988
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967836517466718,
      "compression_loss": 0.0,
      "distillation_loss": 0.08589686453342438,
      "epoch": 5.46,
      "learning_rate": 5.7873884286781615e-06,
      "loss": 0.0875,
      "step": 5746,
      "task_loss": 0.10212301462888718
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967872521361732,
      "compression_loss": 0.0,
      "distillation_loss": 0.032520610839128494,
      "epoch": 5.46,
      "learning_rate": 5.7805724743136445e-06,
      "loss": 0.0341,
      "step": 5747,
      "task_loss": 0.04851553216576576
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796790849837814,
      "compression_loss": 0.0,
      "distillation_loss": 0.029095636680722237,
      "epoch": 5.46,
      "learning_rate": 5.7737600112506925e-06,
      "loss": 0.0268,
      "step": 5748,
      "task_loss": 0.00580623559653759
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967944448525979,
      "compression_loss": 0.0,
      "distillation_loss": 0.13918885588645935,
      "epoch": 5.46,
      "learning_rate": 5.766951040726837e-06,
      "loss": 0.134,
      "step": 5749,
      "task_loss": 0.0872417539358139
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7967980371815284,
      "compression_loss": 0.0,
      "distillation_loss": 0.03975678235292435,
      "epoch": 5.46,
      "learning_rate": 5.760145563978953e-06,
      "loss": 0.0369,
      "step": 5750,
      "task_loss": 0.011658702045679092
    },
    {
      "epoch": 5.46,
      "eval_accuracy": 0.8853211009174312,
      "eval_loss": 0.4336629807949066,
      "eval_runtime": 17.9957,
      "eval_samples_per_second": 48.456,
      "eval_steps_per_second": 6.057,
      "step": 5750
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968016268256093,
      "compression_loss": 0.0,
      "distillation_loss": 0.015128877013921738,
      "epoch": 5.46,
      "learning_rate": 5.753343582243278e-06,
      "loss": 0.014,
      "step": 5751,
      "task_loss": 0.0033825524151325226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968052137858443,
      "compression_loss": 0.0,
      "distillation_loss": 0.021273450925946236,
      "epoch": 5.46,
      "learning_rate": 5.746545096755437e-06,
      "loss": 0.024,
      "step": 5752,
      "task_loss": 0.04804559051990509
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968087980632371,
      "compression_loss": 0.0,
      "distillation_loss": 0.08617687970399857,
      "epoch": 5.46,
      "learning_rate": 5.739750108750408e-06,
      "loss": 0.085,
      "step": 5753,
      "task_loss": 0.07489560544490814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968123796587915,
      "compression_loss": 0.0,
      "distillation_loss": 0.12841589748859406,
      "epoch": 5.46,
      "learning_rate": 5.732958619462533e-06,
      "loss": 0.1363,
      "step": 5754,
      "task_loss": 0.2073012888431549
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968159585735108,
      "compression_loss": 0.0,
      "distillation_loss": 0.03129749745130539,
      "epoch": 5.47,
      "learning_rate": 5.72617063012551e-06,
      "loss": 0.0372,
      "step": 5755,
      "task_loss": 0.09041489660739899
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796819534808399,
      "compression_loss": 0.0,
      "distillation_loss": 0.04228182137012482,
      "epoch": 5.47,
      "learning_rate": 5.719386141972419e-06,
      "loss": 0.0393,
      "step": 5756,
      "task_loss": 0.012629145756363869
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968231083644597,
      "compression_loss": 0.0,
      "distillation_loss": 0.05197533220052719,
      "epoch": 5.47,
      "learning_rate": 5.712605156235695e-06,
      "loss": 0.0484,
      "step": 5757,
      "task_loss": 0.016697099432349205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968266792426966,
      "compression_loss": 0.0,
      "distillation_loss": 0.026457346975803375,
      "epoch": 5.47,
      "learning_rate": 5.7058276741471236e-06,
      "loss": 0.0426,
      "step": 5758,
      "task_loss": 0.187742680311203
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968302474441132,
      "compression_loss": 0.0,
      "distillation_loss": 0.08368387818336487,
      "epoch": 5.47,
      "learning_rate": 5.6990536969378865e-06,
      "loss": 0.0848,
      "step": 5759,
      "task_loss": 0.09534468501806259
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968338129697136,
      "compression_loss": 0.0,
      "distillation_loss": 0.036162346601486206,
      "epoch": 5.47,
      "learning_rate": 5.692283225838493e-06,
      "loss": 0.0487,
      "step": 5760,
      "task_loss": 0.1613989919424057
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796837375820501,
      "compression_loss": 0.0,
      "distillation_loss": 0.05870455503463745,
      "epoch": 5.47,
      "learning_rate": 5.6855162620788435e-06,
      "loss": 0.057,
      "step": 5761,
      "task_loss": 0.04194345325231552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968409359974794,
      "compression_loss": 0.0,
      "distillation_loss": 0.07059012353420258,
      "epoch": 5.47,
      "learning_rate": 5.67875280688818e-06,
      "loss": 0.0739,
      "step": 5762,
      "task_loss": 0.1037897914648056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968444935016523,
      "compression_loss": 0.0,
      "distillation_loss": 0.015868360176682472,
      "epoch": 5.47,
      "learning_rate": 5.671992861495126e-06,
      "loss": 0.0187,
      "step": 5763,
      "task_loss": 0.04378681629896164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968480483340236,
      "compression_loss": 0.0,
      "distillation_loss": 0.02727351151406765,
      "epoch": 5.47,
      "learning_rate": 5.665236427127654e-06,
      "loss": 0.0267,
      "step": 5764,
      "task_loss": 0.021995794028043747
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968516004955968,
      "compression_loss": 0.0,
      "distillation_loss": 0.04122058302164078,
      "epoch": 5.47,
      "learning_rate": 5.658483505013096e-06,
      "loss": 0.0494,
      "step": 5765,
      "task_loss": 0.12348385155200958
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968551499873756,
      "compression_loss": 0.0,
      "distillation_loss": 0.10042732208967209,
      "epoch": 5.48,
      "learning_rate": 5.651734096378164e-06,
      "loss": 0.1009,
      "step": 5766,
      "task_loss": 0.10483403503894806
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968586968103637,
      "compression_loss": 0.0,
      "distillation_loss": 0.062294743955135345,
      "epoch": 5.48,
      "learning_rate": 5.644988202448917e-06,
      "loss": 0.0605,
      "step": 5767,
      "task_loss": 0.04418262839317322
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968622409655649,
      "compression_loss": 0.0,
      "distillation_loss": 0.12338921427726746,
      "epoch": 5.48,
      "learning_rate": 5.638245824450777e-06,
      "loss": 0.1259,
      "step": 5768,
      "task_loss": 0.14866846799850464
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968657824539827,
      "compression_loss": 0.0,
      "distillation_loss": 0.041369691491127014,
      "epoch": 5.48,
      "learning_rate": 5.631506963608518e-06,
      "loss": 0.0491,
      "step": 5769,
      "task_loss": 0.11868831515312195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968693212766209,
      "compression_loss": 0.0,
      "distillation_loss": 0.07022303342819214,
      "epoch": 5.48,
      "learning_rate": 5.624771621146313e-06,
      "loss": 0.0894,
      "step": 5770,
      "task_loss": 0.2618139982223511
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968728574344831,
      "compression_loss": 0.0,
      "distillation_loss": 0.024889115244150162,
      "epoch": 5.48,
      "learning_rate": 5.618039798287652e-06,
      "loss": 0.0231,
      "step": 5771,
      "task_loss": 0.006804602220654488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796876390928573,
      "compression_loss": 0.0,
      "distillation_loss": 0.04423792287707329,
      "epoch": 5.48,
      "learning_rate": 5.6113114962554035e-06,
      "loss": 0.0483,
      "step": 5772,
      "task_loss": 0.08468769490718842
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968799217598944,
      "compression_loss": 0.0,
      "distillation_loss": 0.036272380501031876,
      "epoch": 5.48,
      "learning_rate": 5.6045867162718e-06,
      "loss": 0.0509,
      "step": 5773,
      "task_loss": 0.18288281559944153
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968834499294508,
      "compression_loss": 0.0,
      "distillation_loss": 0.1168503686785698,
      "epoch": 5.48,
      "learning_rate": 5.597865459558435e-06,
      "loss": 0.1163,
      "step": 5774,
      "task_loss": 0.11171647161245346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796886975438246,
      "compression_loss": 0.0,
      "distillation_loss": 0.06244365870952606,
      "epoch": 5.48,
      "learning_rate": 5.591147727336246e-06,
      "loss": 0.0676,
      "step": 5775,
      "task_loss": 0.11397860199213028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968904982872836,
      "compression_loss": 0.0,
      "distillation_loss": 0.027232658118009567,
      "epoch": 5.49,
      "learning_rate": 5.584433520825541e-06,
      "loss": 0.0281,
      "step": 5776,
      "task_loss": 0.03567790612578392
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7968940184775675,
      "compression_loss": 0.0,
      "distillation_loss": 0.08206064254045486,
      "epoch": 5.49,
      "learning_rate": 5.577722841245995e-06,
      "loss": 0.0808,
      "step": 5777,
      "task_loss": 0.06914728134870529
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796897536010101,
      "compression_loss": 0.0,
      "distillation_loss": 0.06064610183238983,
      "epoch": 5.49,
      "learning_rate": 5.571015689816639e-06,
      "loss": 0.0668,
      "step": 5778,
      "task_loss": 0.1217413991689682
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969010508858883,
      "compression_loss": 0.0,
      "distillation_loss": 0.045650772750377655,
      "epoch": 5.49,
      "learning_rate": 5.564312067755856e-06,
      "loss": 0.0601,
      "step": 5779,
      "task_loss": 0.19037003815174103
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969045631059325,
      "compression_loss": 0.0,
      "distillation_loss": 0.07006368786096573,
      "epoch": 5.49,
      "learning_rate": 5.5576119762813795e-06,
      "loss": 0.0666,
      "step": 5780,
      "task_loss": 0.03572140634059906
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969080726712376,
      "compression_loss": 0.0,
      "distillation_loss": 0.02565773017704487,
      "epoch": 5.49,
      "learning_rate": 5.550915416610331e-06,
      "loss": 0.0336,
      "step": 5781,
      "task_loss": 0.10518058389425278
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969115795828073,
      "compression_loss": 0.0,
      "distillation_loss": 0.03800772875547409,
      "epoch": 5.49,
      "learning_rate": 5.544222389959164e-06,
      "loss": 0.0355,
      "step": 5782,
      "task_loss": 0.012635795399546623
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969150838416453,
      "compression_loss": 0.0,
      "distillation_loss": 0.03859638050198555,
      "epoch": 5.49,
      "learning_rate": 5.537532897543695e-06,
      "loss": 0.0359,
      "step": 5783,
      "task_loss": 0.011166783049702644
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969185854487552,
      "compression_loss": 0.0,
      "distillation_loss": 0.05814467743039131,
      "epoch": 5.49,
      "learning_rate": 5.530846940579112e-06,
      "loss": 0.0551,
      "step": 5784,
      "task_loss": 0.028073014691472054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969220844051406,
      "compression_loss": 0.0,
      "distillation_loss": 0.02096547558903694,
      "epoch": 5.49,
      "learning_rate": 5.524164520279948e-06,
      "loss": 0.0194,
      "step": 5785,
      "task_loss": 0.005785791203379631
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969255807118053,
      "compression_loss": 0.0,
      "distillation_loss": 0.03538232296705246,
      "epoch": 5.49,
      "learning_rate": 5.5174856378600895e-06,
      "loss": 0.0328,
      "step": 5786,
      "task_loss": 0.00995618849992752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796929074369753,
      "compression_loss": 0.0,
      "distillation_loss": 0.017818935215473175,
      "epoch": 5.5,
      "learning_rate": 5.510810294532792e-06,
      "loss": 0.0164,
      "step": 5787,
      "task_loss": 0.004038920626044273
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969325653799872,
      "compression_loss": 0.0,
      "distillation_loss": 0.0632765144109726,
      "epoch": 5.5,
      "learning_rate": 5.504138491510674e-06,
      "loss": 0.067,
      "step": 5788,
      "task_loss": 0.1003769114613533
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796936053743512,
      "compression_loss": 0.0,
      "distillation_loss": 0.03711196035146713,
      "epoch": 5.5,
      "learning_rate": 5.497470230005691e-06,
      "loss": 0.044,
      "step": 5789,
      "task_loss": 0.10593129694461823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969395394613306,
      "compression_loss": 0.0,
      "distillation_loss": 0.07631154358386993,
      "epoch": 5.5,
      "learning_rate": 5.490805511229158e-06,
      "loss": 0.0861,
      "step": 5790,
      "task_loss": 0.173743337392807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969430225344468,
      "compression_loss": 0.0,
      "distillation_loss": 0.03230639174580574,
      "epoch": 5.5,
      "learning_rate": 5.484144336391769e-06,
      "loss": 0.0373,
      "step": 5791,
      "task_loss": 0.08267946541309357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969465029638645,
      "compression_loss": 0.0,
      "distillation_loss": 0.031054720282554626,
      "epoch": 5.5,
      "learning_rate": 5.477486706703553e-06,
      "loss": 0.0289,
      "step": 5792,
      "task_loss": 0.009296312928199768
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969499807505873,
      "compression_loss": 0.0,
      "distillation_loss": 0.09885191917419434,
      "epoch": 5.5,
      "learning_rate": 5.470832623373889e-06,
      "loss": 0.1043,
      "step": 5793,
      "task_loss": 0.1528811752796173
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969534558956187,
      "compression_loss": 0.0,
      "distillation_loss": 0.0922151654958725,
      "epoch": 5.5,
      "learning_rate": 5.464182087611538e-06,
      "loss": 0.0877,
      "step": 5794,
      "task_loss": 0.047085706144571304
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969569283999626,
      "compression_loss": 0.0,
      "distillation_loss": 0.037871234118938446,
      "epoch": 5.5,
      "learning_rate": 5.457535100624592e-06,
      "loss": 0.0475,
      "step": 5795,
      "task_loss": 0.133943110704422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969603982646226,
      "compression_loss": 0.0,
      "distillation_loss": 0.1495521366596222,
      "epoch": 5.5,
      "learning_rate": 5.450891663620519e-06,
      "loss": 0.1464,
      "step": 5796,
      "task_loss": 0.11768987029790878
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969638654906023,
      "compression_loss": 0.0,
      "distillation_loss": 0.03055592253804207,
      "epoch": 5.51,
      "learning_rate": 5.444251777806117e-06,
      "loss": 0.0286,
      "step": 5797,
      "task_loss": 0.011272316798567772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969673300789055,
      "compression_loss": 0.0,
      "distillation_loss": 0.1179354190826416,
      "epoch": 5.51,
      "learning_rate": 5.43761544438757e-06,
      "loss": 0.1202,
      "step": 5798,
      "task_loss": 0.14022500813007355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969707920305359,
      "compression_loss": 0.0,
      "distillation_loss": 0.02442678064107895,
      "epoch": 5.51,
      "learning_rate": 5.4309826645703886e-06,
      "loss": 0.0225,
      "step": 5799,
      "task_loss": 0.0055536795407533646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.796974251346497,
      "compression_loss": 0.0,
      "distillation_loss": 0.10044455528259277,
      "epoch": 5.51,
      "learning_rate": 5.424353439559446e-06,
      "loss": 0.1063,
      "step": 5800,
      "task_loss": 0.15924617648124695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969777080277928,
      "compression_loss": 0.0,
      "distillation_loss": 0.028251107782125473,
      "epoch": 5.51,
      "learning_rate": 5.417727770558984e-06,
      "loss": 0.0266,
      "step": 5801,
      "task_loss": 0.012232955545186996
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969811620754266,
      "compression_loss": 0.0,
      "distillation_loss": 0.02132299914956093,
      "epoch": 5.51,
      "learning_rate": 5.4111056587725836e-06,
      "loss": 0.0201,
      "step": 5802,
      "task_loss": 0.009178368374705315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969846134904024,
      "compression_loss": 0.0,
      "distillation_loss": 0.041062891483306885,
      "epoch": 5.51,
      "learning_rate": 5.404487105403172e-06,
      "loss": 0.0377,
      "step": 5803,
      "task_loss": 0.00785096362233162
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969880622737237,
      "compression_loss": 0.0,
      "distillation_loss": 0.027310824021697044,
      "epoch": 5.51,
      "learning_rate": 5.397872111653052e-06,
      "loss": 0.0253,
      "step": 5804,
      "task_loss": 0.006779264658689499
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969915084263942,
      "compression_loss": 0.0,
      "distillation_loss": 0.17357215285301208,
      "epoch": 5.51,
      "learning_rate": 5.3912606787238754e-06,
      "loss": 0.1662,
      "step": 5805,
      "task_loss": 0.09964841604232788
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969949519494176,
      "compression_loss": 0.0,
      "distillation_loss": 0.03900089114904404,
      "epoch": 5.51,
      "learning_rate": 5.384652807816631e-06,
      "loss": 0.0428,
      "step": 5806,
      "task_loss": 0.07701753824949265
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7969983928437976,
      "compression_loss": 0.0,
      "distillation_loss": 0.026013296097517014,
      "epoch": 5.51,
      "learning_rate": 5.378048500131669e-06,
      "loss": 0.0246,
      "step": 5807,
      "task_loss": 0.011500442400574684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970018311105379,
      "compression_loss": 0.0,
      "distillation_loss": 0.020743299275636673,
      "epoch": 5.52,
      "learning_rate": 5.3714477568687025e-06,
      "loss": 0.019,
      "step": 5808,
      "task_loss": 0.0032861437648534775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970052667506422,
      "compression_loss": 0.0,
      "distillation_loss": 0.11350180208683014,
      "epoch": 5.52,
      "learning_rate": 5.3648505792267825e-06,
      "loss": 0.1343,
      "step": 5809,
      "task_loss": 0.3215253949165344
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970086997651141,
      "compression_loss": 0.0,
      "distillation_loss": 0.03125522658228874,
      "epoch": 5.52,
      "learning_rate": 5.358256968404312e-06,
      "loss": 0.0371,
      "step": 5810,
      "task_loss": 0.08946846425533295
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970121301549574,
      "compression_loss": 0.0,
      "distillation_loss": 0.03678523749113083,
      "epoch": 5.52,
      "learning_rate": 5.351666925599067e-06,
      "loss": 0.0436,
      "step": 5811,
      "task_loss": 0.10449925065040588
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970155579211756,
      "compression_loss": 0.0,
      "distillation_loss": 0.013445551507174969,
      "epoch": 5.52,
      "learning_rate": 5.345080452008145e-06,
      "loss": 0.0124,
      "step": 5812,
      "task_loss": 0.002551089972257614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970189830647726,
      "compression_loss": 0.0,
      "distillation_loss": 0.09279379993677139,
      "epoch": 5.52,
      "learning_rate": 5.338497548828025e-06,
      "loss": 0.0901,
      "step": 5813,
      "task_loss": 0.06628113985061646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797022405586752,
      "compression_loss": 0.0,
      "distillation_loss": 0.10782104730606079,
      "epoch": 5.52,
      "learning_rate": 5.331918217254508e-06,
      "loss": 0.1014,
      "step": 5814,
      "task_loss": 0.043450452387332916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970258254881173,
      "compression_loss": 0.0,
      "distillation_loss": 0.022084686905145645,
      "epoch": 5.52,
      "learning_rate": 5.325342458482779e-06,
      "loss": 0.0385,
      "step": 5815,
      "task_loss": 0.18579933047294617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970292427698724,
      "compression_loss": 0.0,
      "distillation_loss": 0.12914645671844482,
      "epoch": 5.52,
      "learning_rate": 5.3187702737073435e-06,
      "loss": 0.1435,
      "step": 5816,
      "task_loss": 0.2728707790374756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797032657433021,
      "compression_loss": 0.0,
      "distillation_loss": 0.03833978995680809,
      "epoch": 5.52,
      "learning_rate": 5.312201664122068e-06,
      "loss": 0.0571,
      "step": 5817,
      "task_loss": 0.22600466012954712
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970360694785666,
      "compression_loss": 0.0,
      "distillation_loss": 0.02015780098736286,
      "epoch": 5.53,
      "learning_rate": 5.305636630920186e-06,
      "loss": 0.0231,
      "step": 5818,
      "task_loss": 0.04994331672787666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797039478907513,
      "compression_loss": 0.0,
      "distillation_loss": 0.12706497311592102,
      "epoch": 5.53,
      "learning_rate": 5.299075175294258e-06,
      "loss": 0.1271,
      "step": 5819,
      "task_loss": 0.1269303411245346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970428857208639,
      "compression_loss": 0.0,
      "distillation_loss": 0.03767715394496918,
      "epoch": 5.53,
      "learning_rate": 5.2925172984361944e-06,
      "loss": 0.0446,
      "step": 5820,
      "task_loss": 0.10687967389822006
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797046289919623,
      "compression_loss": 0.0,
      "distillation_loss": 0.16567179560661316,
      "epoch": 5.53,
      "learning_rate": 5.2859630015372804e-06,
      "loss": 0.1592,
      "step": 5821,
      "task_loss": 0.10095572471618652
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970496915047939,
      "compression_loss": 0.0,
      "distillation_loss": 0.1395283043384552,
      "epoch": 5.53,
      "learning_rate": 5.279412285788138e-06,
      "loss": 0.1345,
      "step": 5822,
      "task_loss": 0.08924393355846405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970530904773803,
      "compression_loss": 0.0,
      "distillation_loss": 0.034282200038433075,
      "epoch": 5.53,
      "learning_rate": 5.2728651523787285e-06,
      "loss": 0.0354,
      "step": 5823,
      "task_loss": 0.045323435217142105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970564868383858,
      "compression_loss": 0.0,
      "distillation_loss": 0.045332517474889755,
      "epoch": 5.53,
      "learning_rate": 5.266321602498361e-06,
      "loss": 0.0573,
      "step": 5824,
      "task_loss": 0.16505207121372223
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970598805888143,
      "compression_loss": 0.0,
      "distillation_loss": 0.036005791276693344,
      "epoch": 5.53,
      "learning_rate": 5.2597816373357226e-06,
      "loss": 0.0365,
      "step": 5825,
      "task_loss": 0.04140361398458481
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970632717296693,
      "compression_loss": 0.0,
      "distillation_loss": 0.06637313961982727,
      "epoch": 5.53,
      "learning_rate": 5.25324525807882e-06,
      "loss": 0.0766,
      "step": 5826,
      "task_loss": 0.16905876994132996
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970666602619545,
      "compression_loss": 0.0,
      "distillation_loss": 0.06400737166404724,
      "epoch": 5.53,
      "learning_rate": 5.246712465915011e-06,
      "loss": 0.0715,
      "step": 5827,
      "task_loss": 0.13885146379470825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970700461866737,
      "compression_loss": 0.0,
      "distillation_loss": 0.07488502562046051,
      "epoch": 5.53,
      "learning_rate": 5.240183262031021e-06,
      "loss": 0.0757,
      "step": 5828,
      "task_loss": 0.08303892612457275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970734295048304,
      "compression_loss": 0.0,
      "distillation_loss": 0.09977598488330841,
      "epoch": 5.54,
      "learning_rate": 5.233657647612899e-06,
      "loss": 0.1021,
      "step": 5829,
      "task_loss": 0.12288139760494232
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970768102174285,
      "compression_loss": 0.0,
      "distillation_loss": 0.030375579372048378,
      "epoch": 5.54,
      "learning_rate": 5.227135623846069e-06,
      "loss": 0.0348,
      "step": 5830,
      "task_loss": 0.07510203868150711
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970801883254715,
      "compression_loss": 0.0,
      "distillation_loss": 0.024371018633246422,
      "epoch": 5.54,
      "learning_rate": 5.220617191915272e-06,
      "loss": 0.0227,
      "step": 5831,
      "task_loss": 0.008041396737098694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970835638299631,
      "compression_loss": 0.0,
      "distillation_loss": 0.02607511542737484,
      "epoch": 5.54,
      "learning_rate": 5.214102353004627e-06,
      "loss": 0.0242,
      "step": 5832,
      "task_loss": 0.0076834335923194885
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970869367319072,
      "compression_loss": 0.0,
      "distillation_loss": 0.13398540019989014,
      "epoch": 5.54,
      "learning_rate": 5.207591108297582e-06,
      "loss": 0.1458,
      "step": 5833,
      "task_loss": 0.2525257170200348
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7970903070323072,
      "compression_loss": 0.0,
      "distillation_loss": 0.0372280478477478,
      "epoch": 5.54,
      "learning_rate": 5.201083458976925e-06,
      "loss": 0.0347,
      "step": 5834,
      "task_loss": 0.011982131749391556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797093674732167,
      "compression_loss": 0.0,
      "distillation_loss": 0.08168113976716995,
      "epoch": 5.54,
      "learning_rate": 5.194579406224817e-06,
      "loss": 0.0761,
      "step": 5835,
      "task_loss": 0.026058735325932503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79709703983249,
      "compression_loss": 0.0,
      "distillation_loss": 0.06715555489063263,
      "epoch": 5.54,
      "learning_rate": 5.188078951222744e-06,
      "loss": 0.0648,
      "step": 5836,
      "task_loss": 0.043633393943309784
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971004023342801,
      "compression_loss": 0.0,
      "distillation_loss": 0.3370836079120636,
      "epoch": 5.54,
      "learning_rate": 5.181582095151538e-06,
      "loss": 0.3367,
      "step": 5837,
      "task_loss": 0.3332991600036621
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971037622385411,
      "compression_loss": 0.0,
      "distillation_loss": 0.016430668532848358,
      "epoch": 5.54,
      "learning_rate": 5.175088839191392e-06,
      "loss": 0.019,
      "step": 5838,
      "task_loss": 0.04250016063451767
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971071195462764,
      "compression_loss": 0.0,
      "distillation_loss": 0.029836127534508705,
      "epoch": 5.55,
      "learning_rate": 5.168599184521841e-06,
      "loss": 0.0381,
      "step": 5839,
      "task_loss": 0.11209283769130707
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971104742584899,
      "compression_loss": 0.0,
      "distillation_loss": 0.0611884780228138,
      "epoch": 5.55,
      "learning_rate": 5.162113132321758e-06,
      "loss": 0.0694,
      "step": 5840,
      "task_loss": 0.14316709339618683
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971138263761851,
      "compression_loss": 0.0,
      "distillation_loss": 0.06297457218170166,
      "epoch": 5.55,
      "learning_rate": 5.155630683769358e-06,
      "loss": 0.0591,
      "step": 5841,
      "task_loss": 0.02457287907600403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971171759003658,
      "compression_loss": 0.0,
      "distillation_loss": 0.04800242558121681,
      "epoch": 5.55,
      "learning_rate": 5.149151840042224e-06,
      "loss": 0.0468,
      "step": 5842,
      "task_loss": 0.035977013409137726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971205228320356,
      "compression_loss": 0.0,
      "distillation_loss": 0.09977774322032928,
      "epoch": 5.55,
      "learning_rate": 5.142676602317259e-06,
      "loss": 0.0959,
      "step": 5843,
      "task_loss": 0.060707636177539825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971238671721983,
      "compression_loss": 0.0,
      "distillation_loss": 0.04978405311703682,
      "epoch": 5.55,
      "learning_rate": 5.1362049717707165e-06,
      "loss": 0.0504,
      "step": 5844,
      "task_loss": 0.055560655891895294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971272089218575,
      "compression_loss": 0.0,
      "distillation_loss": 0.07310109585523605,
      "epoch": 5.55,
      "learning_rate": 5.129736949578215e-06,
      "loss": 0.0786,
      "step": 5845,
      "task_loss": 0.1284613013267517
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797130548082017,
      "compression_loss": 0.0,
      "distillation_loss": 0.029148953035473824,
      "epoch": 5.55,
      "learning_rate": 5.123272536914689e-06,
      "loss": 0.0419,
      "step": 5846,
      "task_loss": 0.1563912034034729
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971338846536803,
      "compression_loss": 0.0,
      "distillation_loss": 0.06112636625766754,
      "epoch": 5.55,
      "learning_rate": 5.116811734954429e-06,
      "loss": 0.0628,
      "step": 5847,
      "task_loss": 0.07760436087846756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971372186378511,
      "compression_loss": 0.0,
      "distillation_loss": 0.087627112865448,
      "epoch": 5.55,
      "learning_rate": 5.1103545448710765e-06,
      "loss": 0.0942,
      "step": 5848,
      "task_loss": 0.15320152044296265
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971405500355332,
      "compression_loss": 0.0,
      "distillation_loss": 0.04075030982494354,
      "epoch": 5.55,
      "learning_rate": 5.103900967837618e-06,
      "loss": 0.0486,
      "step": 5849,
      "task_loss": 0.11899574100971222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971438788477302,
      "compression_loss": 0.0,
      "distillation_loss": 0.12920604646205902,
      "epoch": 5.56,
      "learning_rate": 5.097451005026369e-06,
      "loss": 0.1303,
      "step": 5850,
      "task_loss": 0.13977056741714478
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971472050754459,
      "compression_loss": 0.0,
      "distillation_loss": 0.03929203748703003,
      "epoch": 5.56,
      "learning_rate": 5.091004657608989e-06,
      "loss": 0.0456,
      "step": 5851,
      "task_loss": 0.10265965014696121
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971505287196838,
      "compression_loss": 0.0,
      "distillation_loss": 0.029239453375339508,
      "epoch": 5.56,
      "learning_rate": 5.084561926756501e-06,
      "loss": 0.0311,
      "step": 5852,
      "task_loss": 0.047469861805438995
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971538497814477,
      "compression_loss": 0.0,
      "distillation_loss": 0.014326438307762146,
      "epoch": 5.56,
      "learning_rate": 5.078122813639255e-06,
      "loss": 0.0133,
      "step": 5853,
      "task_loss": 0.004335761070251465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971571682617412,
      "compression_loss": 0.0,
      "distillation_loss": 0.06414780765771866,
      "epoch": 5.56,
      "learning_rate": 5.071687319426946e-06,
      "loss": 0.0762,
      "step": 5854,
      "task_loss": 0.1843860000371933
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971604841615682,
      "compression_loss": 0.0,
      "distillation_loss": 0.033587224781513214,
      "epoch": 5.56,
      "learning_rate": 5.065255445288594e-06,
      "loss": 0.0479,
      "step": 5855,
      "task_loss": 0.1766654998064041
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971637974819321,
      "compression_loss": 0.0,
      "distillation_loss": 0.09691178798675537,
      "epoch": 5.56,
      "learning_rate": 5.058827192392613e-06,
      "loss": 0.0928,
      "step": 5856,
      "task_loss": 0.05549832805991173
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971671082238367,
      "compression_loss": 0.0,
      "distillation_loss": 0.057575710117816925,
      "epoch": 5.56,
      "learning_rate": 5.05240256190671e-06,
      "loss": 0.0648,
      "step": 5857,
      "task_loss": 0.12961743772029877
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971704163882857,
      "compression_loss": 0.0,
      "distillation_loss": 0.08201977610588074,
      "epoch": 5.56,
      "learning_rate": 5.045981554997945e-06,
      "loss": 0.0782,
      "step": 5858,
      "task_loss": 0.043463174253702164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971737219762828,
      "compression_loss": 0.0,
      "distillation_loss": 0.015410145744681358,
      "epoch": 5.56,
      "learning_rate": 5.039564172832733e-06,
      "loss": 0.0147,
      "step": 5859,
      "task_loss": 0.008038915693759918
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971770249888316,
      "compression_loss": 0.0,
      "distillation_loss": 0.022354498505592346,
      "epoch": 5.57,
      "learning_rate": 5.0331504165768236e-06,
      "loss": 0.0289,
      "step": 5860,
      "task_loss": 0.08798445761203766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971803254269358,
      "compression_loss": 0.0,
      "distillation_loss": 0.03102359175682068,
      "epoch": 5.57,
      "learning_rate": 5.026740287395298e-06,
      "loss": 0.0288,
      "step": 5861,
      "task_loss": 0.00922648049890995
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971836232915992,
      "compression_loss": 0.0,
      "distillation_loss": 0.04426976293325424,
      "epoch": 5.57,
      "learning_rate": 5.020333786452589e-06,
      "loss": 0.0411,
      "step": 5862,
      "task_loss": 0.012406604364514351
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971869185838254,
      "compression_loss": 0.0,
      "distillation_loss": 0.026221610605716705,
      "epoch": 5.57,
      "learning_rate": 5.013930914912476e-06,
      "loss": 0.0242,
      "step": 5863,
      "task_loss": 0.00634673610329628
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971902113046181,
      "compression_loss": 0.0,
      "distillation_loss": 0.09950605034828186,
      "epoch": 5.57,
      "learning_rate": 5.007531673938059e-06,
      "loss": 0.1281,
      "step": 5864,
      "task_loss": 0.38576656579971313
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971935014549808,
      "compression_loss": 0.0,
      "distillation_loss": 0.03954896330833435,
      "epoch": 5.57,
      "learning_rate": 5.0011360646917996e-06,
      "loss": 0.043,
      "step": 5865,
      "task_loss": 0.07363829016685486
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7971967890359175,
      "compression_loss": 0.0,
      "distillation_loss": 0.25650808215141296,
      "epoch": 5.57,
      "learning_rate": 4.994744088335496e-06,
      "loss": 0.2497,
      "step": 5866,
      "task_loss": 0.1884750872850418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972000740484316,
      "compression_loss": 0.0,
      "distillation_loss": 0.022899843752384186,
      "epoch": 5.57,
      "learning_rate": 4.9883557460302735e-06,
      "loss": 0.0278,
      "step": 5867,
      "task_loss": 0.07162778824567795
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797203356493527,
      "compression_loss": 0.0,
      "distillation_loss": 0.0954747349023819,
      "epoch": 5.57,
      "learning_rate": 4.9819710389366085e-06,
      "loss": 0.091,
      "step": 5868,
      "task_loss": 0.05076561123132706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972066363722073,
      "compression_loss": 0.0,
      "distillation_loss": 0.026118488982319832,
      "epoch": 5.57,
      "learning_rate": 4.975589968214303e-06,
      "loss": 0.0272,
      "step": 5869,
      "task_loss": 0.0369403176009655
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972099136854761,
      "compression_loss": 0.0,
      "distillation_loss": 0.027349824085831642,
      "epoch": 5.57,
      "learning_rate": 4.969212535022527e-06,
      "loss": 0.0296,
      "step": 5870,
      "task_loss": 0.05007128417491913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972131884343372,
      "compression_loss": 0.0,
      "distillation_loss": 0.03833641856908798,
      "epoch": 5.58,
      "learning_rate": 4.962838740519763e-06,
      "loss": 0.0398,
      "step": 5871,
      "task_loss": 0.052747875452041626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972164606197942,
      "compression_loss": 0.0,
      "distillation_loss": 0.019336001947522163,
      "epoch": 5.58,
      "learning_rate": 4.956468585863835e-06,
      "loss": 0.0207,
      "step": 5872,
      "task_loss": 0.03255251795053482
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972197302428509,
      "compression_loss": 0.0,
      "distillation_loss": 0.02944711409509182,
      "epoch": 5.58,
      "learning_rate": 4.950102072211921e-06,
      "loss": 0.036,
      "step": 5873,
      "task_loss": 0.09525460749864578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972229973045108,
      "compression_loss": 0.0,
      "distillation_loss": 0.01285221055150032,
      "epoch": 5.58,
      "learning_rate": 4.943739200720532e-06,
      "loss": 0.0123,
      "step": 5874,
      "task_loss": 0.006926748901605606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972262618057777,
      "compression_loss": 0.0,
      "distillation_loss": 0.12414573132991791,
      "epoch": 5.58,
      "learning_rate": 4.937379972545508e-06,
      "loss": 0.1184,
      "step": 5875,
      "task_loss": 0.06664574146270752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972295237476553,
      "compression_loss": 0.0,
      "distillation_loss": 0.08365512639284134,
      "epoch": 5.58,
      "learning_rate": 4.9310243888420285e-06,
      "loss": 0.0858,
      "step": 5876,
      "task_loss": 0.1054966151714325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972327831311472,
      "compression_loss": 0.0,
      "distillation_loss": 0.03933534771203995,
      "epoch": 5.58,
      "learning_rate": 4.9246724507646305e-06,
      "loss": 0.0408,
      "step": 5877,
      "task_loss": 0.0535866804420948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972360399572572,
      "compression_loss": 0.0,
      "distillation_loss": 0.015180066227912903,
      "epoch": 5.58,
      "learning_rate": 4.918324159467163e-06,
      "loss": 0.0153,
      "step": 5878,
      "task_loss": 0.016459671780467033
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972392942269888,
      "compression_loss": 0.0,
      "distillation_loss": 0.05211641639471054,
      "epoch": 5.58,
      "learning_rate": 4.911979516102822e-06,
      "loss": 0.0481,
      "step": 5879,
      "task_loss": 0.012342775240540504
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972425459413459,
      "compression_loss": 0.0,
      "distillation_loss": 0.027883626520633698,
      "epoch": 5.58,
      "learning_rate": 4.905638521824155e-06,
      "loss": 0.0401,
      "step": 5880,
      "task_loss": 0.1499495804309845
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797245795101332,
      "compression_loss": 0.0,
      "distillation_loss": 0.10857439041137695,
      "epoch": 5.58,
      "learning_rate": 4.899301177783017e-06,
      "loss": 0.1129,
      "step": 5881,
      "task_loss": 0.15197938680648804
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972490417079509,
      "compression_loss": 0.0,
      "distillation_loss": 0.027678653597831726,
      "epoch": 5.59,
      "learning_rate": 4.892967485130631e-06,
      "loss": 0.0368,
      "step": 5882,
      "task_loss": 0.11922503262758255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972522857622062,
      "compression_loss": 0.0,
      "distillation_loss": 0.02238599956035614,
      "epoch": 5.59,
      "learning_rate": 4.886637445017534e-06,
      "loss": 0.0206,
      "step": 5883,
      "task_loss": 0.0042944010347127914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972555272651016,
      "compression_loss": 0.0,
      "distillation_loss": 0.04405056685209274,
      "epoch": 5.59,
      "learning_rate": 4.880311058593617e-06,
      "loss": 0.0569,
      "step": 5884,
      "task_loss": 0.17205776274204254
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972587662176409,
      "compression_loss": 0.0,
      "distillation_loss": 0.038423389196395874,
      "epoch": 5.59,
      "learning_rate": 4.873988327008094e-06,
      "loss": 0.0371,
      "step": 5885,
      "task_loss": 0.02501399628818035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972620026208276,
      "compression_loss": 0.0,
      "distillation_loss": 0.041545622050762177,
      "epoch": 5.59,
      "learning_rate": 4.867669251409512e-06,
      "loss": 0.0476,
      "step": 5886,
      "task_loss": 0.10194090753793716
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972652364756655,
      "compression_loss": 0.0,
      "distillation_loss": 0.018506459891796112,
      "epoch": 5.59,
      "learning_rate": 4.861353832945778e-06,
      "loss": 0.017,
      "step": 5887,
      "task_loss": 0.003717266023159027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972684677831583,
      "compression_loss": 0.0,
      "distillation_loss": 0.020966939628124237,
      "epoch": 5.59,
      "learning_rate": 4.855042072764107e-06,
      "loss": 0.0301,
      "step": 5888,
      "task_loss": 0.11239578574895859
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972716965443095,
      "compression_loss": 0.0,
      "distillation_loss": 0.024587448686361313,
      "epoch": 5.59,
      "learning_rate": 4.848733972011058e-06,
      "loss": 0.0317,
      "step": 5889,
      "task_loss": 0.09537127614021301
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797274922760123,
      "compression_loss": 0.0,
      "distillation_loss": 0.03539607301354408,
      "epoch": 5.59,
      "learning_rate": 4.842429531832529e-06,
      "loss": 0.0323,
      "step": 5890,
      "task_loss": 0.004601247608661652
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972781464316023,
      "compression_loss": 0.0,
      "distillation_loss": 0.054549772292375565,
      "epoch": 5.59,
      "learning_rate": 4.8361287533737674e-06,
      "loss": 0.0532,
      "step": 5891,
      "task_loss": 0.04105132073163986
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972813675597513,
      "compression_loss": 0.0,
      "distillation_loss": 0.02551000751554966,
      "epoch": 5.6,
      "learning_rate": 4.829831637779322e-06,
      "loss": 0.0288,
      "step": 5892,
      "task_loss": 0.058242082595825195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972845861455735,
      "compression_loss": 0.0,
      "distillation_loss": 0.025515582412481308,
      "epoch": 5.6,
      "learning_rate": 4.823538186193097e-06,
      "loss": 0.0285,
      "step": 5893,
      "task_loss": 0.0552404448390007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972878021900727,
      "compression_loss": 0.0,
      "distillation_loss": 0.027075637131929398,
      "epoch": 5.6,
      "learning_rate": 4.817248399758337e-06,
      "loss": 0.025,
      "step": 5894,
      "task_loss": 0.006751839071512222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972910156942524,
      "compression_loss": 0.0,
      "distillation_loss": 0.1096268743276596,
      "epoch": 5.6,
      "learning_rate": 4.810962279617609e-06,
      "loss": 0.1301,
      "step": 5895,
      "task_loss": 0.31454721093177795
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972942266591165,
      "compression_loss": 0.0,
      "distillation_loss": 0.04903785511851311,
      "epoch": 5.6,
      "learning_rate": 4.804679826912803e-06,
      "loss": 0.0731,
      "step": 5896,
      "task_loss": 0.2901209592819214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7972974350856685,
      "compression_loss": 0.0,
      "distillation_loss": 0.0792558565735817,
      "epoch": 5.6,
      "learning_rate": 4.798401042785177e-06,
      "loss": 0.0772,
      "step": 5897,
      "task_loss": 0.05906623229384422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973006409749123,
      "compression_loss": 0.0,
      "distillation_loss": 0.05788847804069519,
      "epoch": 5.6,
      "learning_rate": 4.792125928375296e-06,
      "loss": 0.0816,
      "step": 5898,
      "task_loss": 0.2952689230442047
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973038443278514,
      "compression_loss": 0.0,
      "distillation_loss": 0.02830476313829422,
      "epoch": 5.6,
      "learning_rate": 4.785854484823052e-06,
      "loss": 0.026,
      "step": 5899,
      "task_loss": 0.00540274940431118
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973070451454894,
      "compression_loss": 0.0,
      "distillation_loss": 0.018592754378914833,
      "epoch": 5.6,
      "learning_rate": 4.779586713267695e-06,
      "loss": 0.0171,
      "step": 5900,
      "task_loss": 0.0032580215483903885
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973102434288303,
      "compression_loss": 0.0,
      "distillation_loss": 0.05436631664633751,
      "epoch": 5.6,
      "learning_rate": 4.7733226148478e-06,
      "loss": 0.0595,
      "step": 5901,
      "task_loss": 0.10618877410888672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973134391788775,
      "compression_loss": 0.0,
      "distillation_loss": 0.01007351279258728,
      "epoch": 5.6,
      "learning_rate": 4.767062190701266e-06,
      "loss": 0.0095,
      "step": 5902,
      "task_loss": 0.004046119749546051
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973166323966349,
      "compression_loss": 0.0,
      "distillation_loss": 0.016038384288549423,
      "epoch": 5.61,
      "learning_rate": 4.760805441965321e-06,
      "loss": 0.0263,
      "step": 5903,
      "task_loss": 0.11890155076980591
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973198230831059,
      "compression_loss": 0.0,
      "distillation_loss": 0.07718884944915771,
      "epoch": 5.61,
      "learning_rate": 4.754552369776547e-06,
      "loss": 0.0741,
      "step": 5904,
      "task_loss": 0.046534955501556396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973230112392944,
      "compression_loss": 0.0,
      "distillation_loss": 0.05940188467502594,
      "epoch": 5.61,
      "learning_rate": 4.748302975270838e-06,
      "loss": 0.062,
      "step": 5905,
      "task_loss": 0.08516646176576614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973261968662041,
      "compression_loss": 0.0,
      "distillation_loss": 0.0312301404774189,
      "epoch": 5.61,
      "learning_rate": 4.7420572595834185e-06,
      "loss": 0.0287,
      "step": 5906,
      "task_loss": 0.006120791658759117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973293799648385,
      "compression_loss": 0.0,
      "distillation_loss": 0.013178318738937378,
      "epoch": 5.61,
      "learning_rate": 4.735815223848864e-06,
      "loss": 0.0122,
      "step": 5907,
      "task_loss": 0.003270508721470833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973325605362014,
      "compression_loss": 0.0,
      "distillation_loss": 0.05070923641324043,
      "epoch": 5.61,
      "learning_rate": 4.7295768692010715e-06,
      "loss": 0.0471,
      "step": 5908,
      "task_loss": 0.014933045953512192
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973357385812966,
      "compression_loss": 0.0,
      "distillation_loss": 0.030595645308494568,
      "epoch": 5.61,
      "learning_rate": 4.723342196773267e-06,
      "loss": 0.0302,
      "step": 5909,
      "task_loss": 0.026463741436600685
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973389141011274,
      "compression_loss": 0.0,
      "distillation_loss": 0.03401756286621094,
      "epoch": 5.61,
      "learning_rate": 4.7171112076979965e-06,
      "loss": 0.0317,
      "step": 5910,
      "task_loss": 0.010913487523794174
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797342087096698,
      "compression_loss": 0.0,
      "distillation_loss": 0.05812390148639679,
      "epoch": 5.61,
      "learning_rate": 4.710883903107166e-06,
      "loss": 0.0634,
      "step": 5911,
      "task_loss": 0.11132801324129105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973452575690118,
      "compression_loss": 0.0,
      "distillation_loss": 0.12190339714288712,
      "epoch": 5.61,
      "learning_rate": 4.70466028413199e-06,
      "loss": 0.1407,
      "step": 5912,
      "task_loss": 0.3099902868270874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973484255190724,
      "compression_loss": 0.0,
      "distillation_loss": 0.01616412214934826,
      "epoch": 5.62,
      "learning_rate": 4.6984403519030076e-06,
      "loss": 0.0148,
      "step": 5913,
      "task_loss": 0.002981981262564659
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973515909478835,
      "compression_loss": 0.0,
      "distillation_loss": 0.0333518460392952,
      "epoch": 5.62,
      "learning_rate": 4.692224107550117e-06,
      "loss": 0.0307,
      "step": 5914,
      "task_loss": 0.007058465853333473
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973547538564489,
      "compression_loss": 0.0,
      "distillation_loss": 0.008787909522652626,
      "epoch": 5.62,
      "learning_rate": 4.686011552202518e-06,
      "loss": 0.0082,
      "step": 5915,
      "task_loss": 0.003183361142873764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973579142457724,
      "compression_loss": 0.0,
      "distillation_loss": 0.04432186111807823,
      "epoch": 5.62,
      "learning_rate": 4.679802686988749e-06,
      "loss": 0.0479,
      "step": 5916,
      "task_loss": 0.0803055539727211
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973610721168575,
      "compression_loss": 0.0,
      "distillation_loss": 0.018226707354187965,
      "epoch": 5.62,
      "learning_rate": 4.673597513036684e-06,
      "loss": 0.0243,
      "step": 5917,
      "task_loss": 0.0785667896270752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973642274707078,
      "compression_loss": 0.0,
      "distillation_loss": 0.033110909163951874,
      "epoch": 5.62,
      "learning_rate": 4.667396031473534e-06,
      "loss": 0.0325,
      "step": 5918,
      "task_loss": 0.02661288157105446
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973673803083271,
      "compression_loss": 0.0,
      "distillation_loss": 0.06958150863647461,
      "epoch": 5.62,
      "learning_rate": 4.661198243425813e-06,
      "loss": 0.0723,
      "step": 5919,
      "task_loss": 0.0964449942111969
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973705306307192,
      "compression_loss": 0.0,
      "distillation_loss": 0.04093363136053085,
      "epoch": 5.62,
      "learning_rate": 4.655004150019379e-06,
      "loss": 0.0406,
      "step": 5920,
      "task_loss": 0.03757309168577194
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973736784388875,
      "compression_loss": 0.0,
      "distillation_loss": 0.0374174565076828,
      "epoch": 5.62,
      "learning_rate": 4.648813752379433e-06,
      "loss": 0.0439,
      "step": 5921,
      "task_loss": 0.10244208574295044
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973768237338359,
      "compression_loss": 0.0,
      "distillation_loss": 0.017319556325674057,
      "epoch": 5.62,
      "learning_rate": 4.642627051630477e-06,
      "loss": 0.0167,
      "step": 5922,
      "task_loss": 0.011339414864778519
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973799665165681,
      "compression_loss": 0.0,
      "distillation_loss": 0.028723616153001785,
      "epoch": 5.62,
      "learning_rate": 4.636444048896355e-06,
      "loss": 0.0353,
      "step": 5923,
      "task_loss": 0.09466679394245148
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973831067880875,
      "compression_loss": 0.0,
      "distillation_loss": 0.08423987776041031,
      "epoch": 5.63,
      "learning_rate": 4.63026474530025e-06,
      "loss": 0.0834,
      "step": 5924,
      "task_loss": 0.07615604996681213
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973862445493982,
      "compression_loss": 0.0,
      "distillation_loss": 0.026835883036255836,
      "epoch": 5.63,
      "learning_rate": 4.624089141964649e-06,
      "loss": 0.0268,
      "step": 5925,
      "task_loss": 0.025994790717959404
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973893798015036,
      "compression_loss": 0.0,
      "distillation_loss": 0.05702338367700577,
      "epoch": 5.63,
      "learning_rate": 4.617917240011394e-06,
      "loss": 0.0545,
      "step": 5926,
      "task_loss": 0.032026879489421844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973925125454074,
      "compression_loss": 0.0,
      "distillation_loss": 0.026394031941890717,
      "epoch": 5.63,
      "learning_rate": 4.611749040561625e-06,
      "loss": 0.0242,
      "step": 5927,
      "task_loss": 0.004607599228620529
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973956427821134,
      "compression_loss": 0.0,
      "distillation_loss": 0.1339588314294815,
      "epoch": 5.63,
      "learning_rate": 4.6055845447358415e-06,
      "loss": 0.1297,
      "step": 5928,
      "task_loss": 0.09128076583147049
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7973987705126252,
      "compression_loss": 0.0,
      "distillation_loss": 0.06346623599529266,
      "epoch": 5.63,
      "learning_rate": 4.599423753653845e-06,
      "loss": 0.0697,
      "step": 5929,
      "task_loss": 0.12597879767417908
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974018957379464,
      "compression_loss": 0.0,
      "distillation_loss": 0.06505750119686127,
      "epoch": 5.63,
      "learning_rate": 4.593266668434767e-06,
      "loss": 0.0716,
      "step": 5930,
      "task_loss": 0.13001671433448792
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974050184590808,
      "compression_loss": 0.0,
      "distillation_loss": 0.05653805285692215,
      "epoch": 5.63,
      "learning_rate": 4.587113290197087e-06,
      "loss": 0.063,
      "step": 5931,
      "task_loss": 0.12097219377756119
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974081386770322,
      "compression_loss": 0.0,
      "distillation_loss": 0.07613807171583176,
      "epoch": 5.63,
      "learning_rate": 4.580963620058587e-06,
      "loss": 0.0749,
      "step": 5932,
      "task_loss": 0.06361458450555801
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974112563928041,
      "compression_loss": 0.0,
      "distillation_loss": 0.16531898081302643,
      "epoch": 5.63,
      "learning_rate": 4.5748176591363795e-06,
      "loss": 0.1618,
      "step": 5933,
      "task_loss": 0.129730224609375
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974143716074001,
      "compression_loss": 0.0,
      "distillation_loss": 0.03830117732286453,
      "epoch": 5.64,
      "learning_rate": 4.56867540854691e-06,
      "loss": 0.0433,
      "step": 5934,
      "task_loss": 0.088396355509758
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974174843218241,
      "compression_loss": 0.0,
      "distillation_loss": 0.0483672171831131,
      "epoch": 5.64,
      "learning_rate": 4.562536869405959e-06,
      "loss": 0.0454,
      "step": 5935,
      "task_loss": 0.01907249167561531
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974205945370797,
      "compression_loss": 0.0,
      "distillation_loss": 0.029438257217407227,
      "epoch": 5.64,
      "learning_rate": 4.556402042828611e-06,
      "loss": 0.0342,
      "step": 5936,
      "task_loss": 0.07672730088233948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974237022541706,
      "compression_loss": 0.0,
      "distillation_loss": 0.024926815181970596,
      "epoch": 5.64,
      "learning_rate": 4.550270929929288e-06,
      "loss": 0.0329,
      "step": 5937,
      "task_loss": 0.10498930513858795
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974268074741003,
      "compression_loss": 0.0,
      "distillation_loss": 0.10737212002277374,
      "epoch": 5.64,
      "learning_rate": 4.54414353182174e-06,
      "loss": 0.103,
      "step": 5938,
      "task_loss": 0.06395427882671356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974299101978728,
      "compression_loss": 0.0,
      "distillation_loss": 0.10752628743648529,
      "epoch": 5.64,
      "learning_rate": 4.538019849619035e-06,
      "loss": 0.0984,
      "step": 5939,
      "task_loss": 0.01625998131930828
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974330104264915,
      "compression_loss": 0.0,
      "distillation_loss": 0.033906545490026474,
      "epoch": 5.64,
      "learning_rate": 4.531899884433574e-06,
      "loss": 0.034,
      "step": 5940,
      "task_loss": 0.035197898745536804
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974361081609602,
      "compression_loss": 0.0,
      "distillation_loss": 0.10264087468385696,
      "epoch": 5.64,
      "learning_rate": 4.525783637377065e-06,
      "loss": 0.0959,
      "step": 5941,
      "task_loss": 0.03510020300745964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974392034022827,
      "compression_loss": 0.0,
      "distillation_loss": 0.021447885781526566,
      "epoch": 5.64,
      "learning_rate": 4.519671109560567e-06,
      "loss": 0.0358,
      "step": 5942,
      "task_loss": 0.16513197124004364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974422961514624,
      "compression_loss": 0.0,
      "distillation_loss": 0.12002286314964294,
      "epoch": 5.64,
      "learning_rate": 4.5135623020944485e-06,
      "loss": 0.1311,
      "step": 5943,
      "task_loss": 0.23128020763397217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974453864095034,
      "compression_loss": 0.0,
      "distillation_loss": 0.04369574412703514,
      "epoch": 5.64,
      "learning_rate": 4.507457216088396e-06,
      "loss": 0.0538,
      "step": 5944,
      "task_loss": 0.14453351497650146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974484741774089,
      "compression_loss": 0.0,
      "distillation_loss": 0.08547208458185196,
      "epoch": 5.65,
      "learning_rate": 4.501355852651443e-06,
      "loss": 0.0862,
      "step": 5945,
      "task_loss": 0.09284761548042297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974515594561828,
      "compression_loss": 0.0,
      "distillation_loss": 0.053911492228507996,
      "epoch": 5.65,
      "learning_rate": 4.495258212891918e-06,
      "loss": 0.058,
      "step": 5946,
      "task_loss": 0.09494040161371231
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974546422468289,
      "compression_loss": 0.0,
      "distillation_loss": 0.0476272851228714,
      "epoch": 5.65,
      "learning_rate": 4.489164297917492e-06,
      "loss": 0.0436,
      "step": 5947,
      "task_loss": 0.00767885148525238
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974577225503507,
      "compression_loss": 0.0,
      "distillation_loss": 0.035985067486763,
      "epoch": 5.65,
      "learning_rate": 4.483074108835145e-06,
      "loss": 0.057,
      "step": 5948,
      "task_loss": 0.24564029276371002
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797460800367752,
      "compression_loss": 0.0,
      "distillation_loss": 0.09636256098747253,
      "epoch": 5.65,
      "learning_rate": 4.476987646751205e-06,
      "loss": 0.1025,
      "step": 5949,
      "task_loss": 0.15778429806232452
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974638757000365,
      "compression_loss": 0.0,
      "distillation_loss": 0.07263234257698059,
      "epoch": 5.65,
      "learning_rate": 4.470904912771298e-06,
      "loss": 0.0894,
      "step": 5950,
      "task_loss": 0.24070273339748383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974669485482078,
      "compression_loss": 0.0,
      "distillation_loss": 0.024517526850104332,
      "epoch": 5.65,
      "learning_rate": 4.46482590800037e-06,
      "loss": 0.0281,
      "step": 5951,
      "task_loss": 0.05985688790678978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974700189132695,
      "compression_loss": 0.0,
      "distillation_loss": 0.06554505228996277,
      "epoch": 5.65,
      "learning_rate": 4.458750633542727e-06,
      "loss": 0.0605,
      "step": 5952,
      "task_loss": 0.015469128265976906
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974730867962254,
      "compression_loss": 0.0,
      "distillation_loss": 0.022991040721535683,
      "epoch": 5.65,
      "learning_rate": 4.45267909050196e-06,
      "loss": 0.0213,
      "step": 5953,
      "task_loss": 0.005838258191943169
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974761521980792,
      "compression_loss": 0.0,
      "distillation_loss": 0.08440842479467392,
      "epoch": 5.65,
      "learning_rate": 4.446611279980992e-06,
      "loss": 0.0993,
      "step": 5954,
      "task_loss": 0.23307430744171143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974792151198347,
      "compression_loss": 0.0,
      "distillation_loss": 0.04542703554034233,
      "epoch": 5.66,
      "learning_rate": 4.440547203082065e-06,
      "loss": 0.0493,
      "step": 5955,
      "task_loss": 0.08394621312618256
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974822755624952,
      "compression_loss": 0.0,
      "distillation_loss": 0.02426799014210701,
      "epoch": 5.66,
      "learning_rate": 4.434486860906761e-06,
      "loss": 0.0256,
      "step": 5956,
      "task_loss": 0.03762784227728844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974853335270646,
      "compression_loss": 0.0,
      "distillation_loss": 0.02789018675684929,
      "epoch": 5.66,
      "learning_rate": 4.4284302545559624e-06,
      "loss": 0.0261,
      "step": 5957,
      "task_loss": 0.010452285408973694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974883890145468,
      "compression_loss": 0.0,
      "distillation_loss": 0.08330793678760529,
      "epoch": 5.66,
      "learning_rate": 4.422377385129878e-06,
      "loss": 0.0919,
      "step": 5958,
      "task_loss": 0.1694803088903427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974914420259451,
      "compression_loss": 0.0,
      "distillation_loss": 0.055434923619031906,
      "epoch": 5.66,
      "learning_rate": 4.416328253728041e-06,
      "loss": 0.0574,
      "step": 5959,
      "task_loss": 0.07511811703443527
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974944925622635,
      "compression_loss": 0.0,
      "distillation_loss": 0.022852642461657524,
      "epoch": 5.66,
      "learning_rate": 4.410282861449317e-06,
      "loss": 0.0288,
      "step": 5960,
      "task_loss": 0.08201512694358826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7974975406245054,
      "compression_loss": 0.0,
      "distillation_loss": 0.023079611361026764,
      "epoch": 5.66,
      "learning_rate": 4.404241209391874e-06,
      "loss": 0.0443,
      "step": 5961,
      "task_loss": 0.23562346398830414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975005862136747,
      "compression_loss": 0.0,
      "distillation_loss": 0.03505755960941315,
      "epoch": 5.66,
      "learning_rate": 4.398203298653195e-06,
      "loss": 0.0437,
      "step": 5962,
      "task_loss": 0.12132131308317184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797503629330775,
      "compression_loss": 0.0,
      "distillation_loss": 0.036127813160419464,
      "epoch": 5.66,
      "learning_rate": 4.392169130330115e-06,
      "loss": 0.0389,
      "step": 5963,
      "task_loss": 0.06360867619514465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79750666997681,
      "compression_loss": 0.0,
      "distillation_loss": 0.03919731080532074,
      "epoch": 5.66,
      "learning_rate": 4.386138705518761e-06,
      "loss": 0.0428,
      "step": 5964,
      "task_loss": 0.07522215694189072
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975097081527833,
      "compression_loss": 0.0,
      "distillation_loss": 0.024520831182599068,
      "epoch": 5.66,
      "learning_rate": 4.380112025314581e-06,
      "loss": 0.0316,
      "step": 5965,
      "task_loss": 0.09532146900892258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975127438596987,
      "compression_loss": 0.0,
      "distillation_loss": 0.04087524488568306,
      "epoch": 5.67,
      "learning_rate": 4.374089090812367e-06,
      "loss": 0.0448,
      "step": 5966,
      "task_loss": 0.0800790786743164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975157770985598,
      "compression_loss": 0.0,
      "distillation_loss": 0.018016282469034195,
      "epoch": 5.67,
      "learning_rate": 4.368069903106203e-06,
      "loss": 0.0224,
      "step": 5967,
      "task_loss": 0.06212467700242996
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975188078703704,
      "compression_loss": 0.0,
      "distillation_loss": 0.02099071443080902,
      "epoch": 5.67,
      "learning_rate": 4.3620544632894996e-06,
      "loss": 0.0255,
      "step": 5968,
      "task_loss": 0.0659455806016922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797521836176134,
      "compression_loss": 0.0,
      "distillation_loss": 0.03363244980573654,
      "epoch": 5.67,
      "learning_rate": 4.3560427724549965e-06,
      "loss": 0.0392,
      "step": 5969,
      "task_loss": 0.08905819803476334
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975248620168545,
      "compression_loss": 0.0,
      "distillation_loss": 0.05440010130405426,
      "epoch": 5.67,
      "learning_rate": 4.350034831694752e-06,
      "loss": 0.0709,
      "step": 5970,
      "task_loss": 0.21891024708747864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975278853935354,
      "compression_loss": 0.0,
      "distillation_loss": 0.0742577612400055,
      "epoch": 5.67,
      "learning_rate": 4.344030642100133e-06,
      "loss": 0.0727,
      "step": 5971,
      "task_loss": 0.05856480449438095
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975309063071804,
      "compression_loss": 0.0,
      "distillation_loss": 0.17016083002090454,
      "epoch": 5.67,
      "learning_rate": 4.33803020476182e-06,
      "loss": 0.1682,
      "step": 5972,
      "task_loss": 0.15005451440811157
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975339247587933,
      "compression_loss": 0.0,
      "distillation_loss": 0.03252163529396057,
      "epoch": 5.67,
      "learning_rate": 4.3320335207698376e-06,
      "loss": 0.03,
      "step": 5973,
      "task_loss": 0.007399743422865868
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975369407493776,
      "compression_loss": 0.0,
      "distillation_loss": 0.022539321333169937,
      "epoch": 5.67,
      "learning_rate": 4.326040591213501e-06,
      "loss": 0.0209,
      "step": 5974,
      "task_loss": 0.006453389301896095
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975399542799372,
      "compression_loss": 0.0,
      "distillation_loss": 0.15156593918800354,
      "epoch": 5.67,
      "learning_rate": 4.320051417181453e-06,
      "loss": 0.1463,
      "step": 5975,
      "task_loss": 0.09853891283273697
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975429653514756,
      "compression_loss": 0.0,
      "distillation_loss": 0.04325627535581589,
      "epoch": 5.68,
      "learning_rate": 4.314065999761668e-06,
      "loss": 0.0558,
      "step": 5976,
      "task_loss": 0.1683274209499359
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975459739649966,
      "compression_loss": 0.0,
      "distillation_loss": 0.019812000915408134,
      "epoch": 5.68,
      "learning_rate": 4.308084340041413e-06,
      "loss": 0.0182,
      "step": 5977,
      "task_loss": 0.003462914377450943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975489801215038,
      "compression_loss": 0.0,
      "distillation_loss": 0.12688830494880676,
      "epoch": 5.68,
      "learning_rate": 4.302106439107298e-06,
      "loss": 0.1374,
      "step": 5978,
      "task_loss": 0.23153865337371826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975519838220009,
      "compression_loss": 0.0,
      "distillation_loss": 0.0461309477686882,
      "epoch": 5.68,
      "learning_rate": 4.296132298045222e-06,
      "loss": 0.0423,
      "step": 5979,
      "task_loss": 0.007924774661660194
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975549850674917,
      "compression_loss": 0.0,
      "distillation_loss": 0.08937639743089676,
      "epoch": 5.68,
      "learning_rate": 4.29016191794043e-06,
      "loss": 0.0852,
      "step": 5980,
      "task_loss": 0.04732952266931534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975579838589797,
      "compression_loss": 0.0,
      "distillation_loss": 0.017793424427509308,
      "epoch": 5.68,
      "learning_rate": 4.284195299877469e-06,
      "loss": 0.0164,
      "step": 5981,
      "task_loss": 0.003985332325100899
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975609801974687,
      "compression_loss": 0.0,
      "distillation_loss": 0.04416952282190323,
      "epoch": 5.68,
      "learning_rate": 4.278232444940192e-06,
      "loss": 0.041,
      "step": 5982,
      "task_loss": 0.01279473677277565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975639740839623,
      "compression_loss": 0.0,
      "distillation_loss": 0.10886112600564957,
      "epoch": 5.68,
      "learning_rate": 4.272273354211795e-06,
      "loss": 0.1035,
      "step": 5983,
      "task_loss": 0.05517327040433884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975669655194643,
      "compression_loss": 0.0,
      "distillation_loss": 0.02088319882750511,
      "epoch": 5.68,
      "learning_rate": 4.266318028774768e-06,
      "loss": 0.0191,
      "step": 5984,
      "task_loss": 0.00327393040060997
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975699545049783,
      "compression_loss": 0.0,
      "distillation_loss": 0.038873784244060516,
      "epoch": 5.68,
      "learning_rate": 4.260366469710919e-06,
      "loss": 0.0364,
      "step": 5985,
      "task_loss": 0.013767654076218605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975729410415079,
      "compression_loss": 0.0,
      "distillation_loss": 0.014923624694347382,
      "epoch": 5.68,
      "learning_rate": 4.254418678101385e-06,
      "loss": 0.0138,
      "step": 5986,
      "task_loss": 0.0034323427826166153
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975759251300569,
      "compression_loss": 0.0,
      "distillation_loss": 0.011618406511843204,
      "epoch": 5.69,
      "learning_rate": 4.248474655026618e-06,
      "loss": 0.011,
      "step": 5987,
      "task_loss": 0.005057178437709808
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797578906771629,
      "compression_loss": 0.0,
      "distillation_loss": 0.03404705598950386,
      "epoch": 5.69,
      "learning_rate": 4.242534401566367e-06,
      "loss": 0.052,
      "step": 5988,
      "task_loss": 0.2138359099626541
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975818859672279,
      "compression_loss": 0.0,
      "distillation_loss": 0.02775607816874981,
      "epoch": 5.69,
      "learning_rate": 4.236597918799709e-06,
      "loss": 0.0253,
      "step": 5989,
      "task_loss": 0.0035394616425037384
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975848627178571,
      "compression_loss": 0.0,
      "distillation_loss": 0.029923155903816223,
      "epoch": 5.69,
      "learning_rate": 4.23066520780504e-06,
      "loss": 0.0275,
      "step": 5990,
      "task_loss": 0.005590047687292099
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975878370245204,
      "compression_loss": 0.0,
      "distillation_loss": 0.03592822700738907,
      "epoch": 5.69,
      "learning_rate": 4.224736269660062e-06,
      "loss": 0.0419,
      "step": 5991,
      "task_loss": 0.09543702751398087
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975908088882216,
      "compression_loss": 0.0,
      "distillation_loss": 0.040161147713661194,
      "epoch": 5.69,
      "learning_rate": 4.218811105441789e-06,
      "loss": 0.0431,
      "step": 5992,
      "task_loss": 0.06955252587795258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975937783099643,
      "compression_loss": 0.0,
      "distillation_loss": 0.013323452323675156,
      "epoch": 5.69,
      "learning_rate": 4.21288971622657e-06,
      "loss": 0.0177,
      "step": 5993,
      "task_loss": 0.056832775473594666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797596745290752,
      "compression_loss": 0.0,
      "distillation_loss": 0.05237278342247009,
      "epoch": 5.69,
      "learning_rate": 4.206972103090037e-06,
      "loss": 0.0474,
      "step": 5994,
      "task_loss": 0.0029510650783777237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7975997098315886,
      "compression_loss": 0.0,
      "distillation_loss": 0.09177492558956146,
      "epoch": 5.69,
      "learning_rate": 4.201058267107169e-06,
      "loss": 0.0933,
      "step": 5995,
      "task_loss": 0.10701367259025574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976026719334777,
      "compression_loss": 0.0,
      "distillation_loss": 0.034797269850969315,
      "epoch": 5.69,
      "learning_rate": 4.195148209352232e-06,
      "loss": 0.0439,
      "step": 5996,
      "task_loss": 0.12537652254104614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976056315974229,
      "compression_loss": 0.0,
      "distillation_loss": 0.010959111154079437,
      "epoch": 5.7,
      "learning_rate": 4.189241930898824e-06,
      "loss": 0.0103,
      "step": 5997,
      "task_loss": 0.00464300811290741
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976085888244282,
      "compression_loss": 0.0,
      "distillation_loss": 0.027968231588602066,
      "epoch": 5.7,
      "learning_rate": 4.183339432819844e-06,
      "loss": 0.0294,
      "step": 5998,
      "task_loss": 0.04234401881694794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976115436154969,
      "compression_loss": 0.0,
      "distillation_loss": 0.06177544593811035,
      "epoch": 5.7,
      "learning_rate": 4.177440716187506e-06,
      "loss": 0.058,
      "step": 5999,
      "task_loss": 0.024354927241802216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976144959716329,
      "compression_loss": 0.0,
      "distillation_loss": 0.024940505623817444,
      "epoch": 5.7,
      "learning_rate": 4.17154578207335e-06,
      "loss": 0.0233,
      "step": 6000,
      "task_loss": 0.008159097284078598
    },
    {
      "epoch": 5.7,
      "eval_accuracy": 0.8830275229357798,
      "eval_loss": 0.45090019702911377,
      "eval_runtime": 18.0915,
      "eval_samples_per_second": 48.199,
      "eval_steps_per_second": 6.025,
      "step": 6000
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976174458938398,
      "compression_loss": 0.0,
      "distillation_loss": 0.0402870811522007,
      "epoch": 5.7,
      "learning_rate": 4.165654631548213e-06,
      "loss": 0.0397,
      "step": 6001,
      "task_loss": 0.03429171442985535
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976203933831213,
      "compression_loss": 0.0,
      "distillation_loss": 0.01571035385131836,
      "epoch": 5.7,
      "learning_rate": 4.159767265682243e-06,
      "loss": 0.0147,
      "step": 6002,
      "task_loss": 0.0054051876068115234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976233384404812,
      "compression_loss": 0.0,
      "distillation_loss": 0.033142976462841034,
      "epoch": 5.7,
      "learning_rate": 4.153883685544921e-06,
      "loss": 0.0392,
      "step": 6003,
      "task_loss": 0.09410811960697174
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976262810669229,
      "compression_loss": 0.0,
      "distillation_loss": 0.03702503442764282,
      "epoch": 5.7,
      "learning_rate": 4.1480038922050295e-06,
      "loss": 0.0537,
      "step": 6004,
      "task_loss": 0.20370852947235107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976292212634503,
      "compression_loss": 0.0,
      "distillation_loss": 0.03896544128656387,
      "epoch": 5.7,
      "learning_rate": 4.142127886730654e-06,
      "loss": 0.0421,
      "step": 6005,
      "task_loss": 0.0707533061504364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976321590310671,
      "compression_loss": 0.0,
      "distillation_loss": 0.02081841602921486,
      "epoch": 5.7,
      "learning_rate": 4.136255670189193e-06,
      "loss": 0.0193,
      "step": 6006,
      "task_loss": 0.005161428824067116
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976350943707768,
      "compression_loss": 0.0,
      "distillation_loss": 0.12577922642230988,
      "epoch": 5.7,
      "learning_rate": 4.130387243647377e-06,
      "loss": 0.1272,
      "step": 6007,
      "task_loss": 0.1395563781261444
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976380272835834,
      "compression_loss": 0.0,
      "distillation_loss": 0.019099004566669464,
      "epoch": 5.71,
      "learning_rate": 4.12452260817123e-06,
      "loss": 0.0204,
      "step": 6008,
      "task_loss": 0.03233078494668007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976409577704902,
      "compression_loss": 0.0,
      "distillation_loss": 0.04859714210033417,
      "epoch": 5.71,
      "learning_rate": 4.118661764826079e-06,
      "loss": 0.051,
      "step": 6009,
      "task_loss": 0.07280280441045761
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976438858325011,
      "compression_loss": 0.0,
      "distillation_loss": 0.018262770026922226,
      "epoch": 5.71,
      "learning_rate": 4.112804714676594e-06,
      "loss": 0.0169,
      "step": 6010,
      "task_loss": 0.004157818853855133
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976468114706198,
      "compression_loss": 0.0,
      "distillation_loss": 0.047912146896123886,
      "epoch": 5.71,
      "learning_rate": 4.106951458786717e-06,
      "loss": 0.0471,
      "step": 6011,
      "task_loss": 0.04020364210009575
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976497346858499,
      "compression_loss": 0.0,
      "distillation_loss": 0.02587919309735298,
      "epoch": 5.71,
      "learning_rate": 4.101101998219737e-06,
      "loss": 0.0394,
      "step": 6012,
      "task_loss": 0.1613081842660904
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976526554791951,
      "compression_loss": 0.0,
      "distillation_loss": 0.025433223694562912,
      "epoch": 5.71,
      "learning_rate": 4.095256334038222e-06,
      "loss": 0.024,
      "step": 6013,
      "task_loss": 0.010708844289183617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976555738516591,
      "compression_loss": 0.0,
      "distillation_loss": 0.025807831436395645,
      "epoch": 5.71,
      "learning_rate": 4.089414467304078e-06,
      "loss": 0.028,
      "step": 6014,
      "task_loss": 0.0474732369184494
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976584898042456,
      "compression_loss": 0.0,
      "distillation_loss": 0.022514909505844116,
      "epoch": 5.71,
      "learning_rate": 4.083576399078504e-06,
      "loss": 0.0324,
      "step": 6015,
      "task_loss": 0.12099795788526535
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976614033379582,
      "compression_loss": 0.0,
      "distillation_loss": 0.013821316882967949,
      "epoch": 5.71,
      "learning_rate": 4.077742130422005e-06,
      "loss": 0.0127,
      "step": 6016,
      "task_loss": 0.002670343965291977
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976643144538006,
      "compression_loss": 0.0,
      "distillation_loss": 0.0412713922560215,
      "epoch": 5.71,
      "learning_rate": 4.071911662394418e-06,
      "loss": 0.0377,
      "step": 6017,
      "task_loss": 0.005630454048514366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976672231527766,
      "compression_loss": 0.0,
      "distillation_loss": 0.07258160412311554,
      "epoch": 5.72,
      "learning_rate": 4.066084996054867e-06,
      "loss": 0.0764,
      "step": 6018,
      "task_loss": 0.11061573028564453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976701294358898,
      "compression_loss": 0.0,
      "distillation_loss": 0.04702713340520859,
      "epoch": 5.72,
      "learning_rate": 4.060262132461795e-06,
      "loss": 0.0465,
      "step": 6019,
      "task_loss": 0.041665218770504
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976730333041439,
      "compression_loss": 0.0,
      "distillation_loss": 0.04431241750717163,
      "epoch": 5.72,
      "learning_rate": 4.054443072672942e-06,
      "loss": 0.042,
      "step": 6020,
      "task_loss": 0.020761430263519287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976759347585425,
      "compression_loss": 0.0,
      "distillation_loss": 0.02348828688263893,
      "epoch": 5.72,
      "learning_rate": 4.048627817745393e-06,
      "loss": 0.0219,
      "step": 6021,
      "task_loss": 0.007251596078276634
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976788338000894,
      "compression_loss": 0.0,
      "distillation_loss": 0.015367396175861359,
      "epoch": 5.72,
      "learning_rate": 4.042816368735502e-06,
      "loss": 0.0143,
      "step": 6022,
      "task_loss": 0.004556819796562195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976817304297882,
      "compression_loss": 0.0,
      "distillation_loss": 0.08127819746732712,
      "epoch": 5.72,
      "learning_rate": 4.037008726698943e-06,
      "loss": 0.0863,
      "step": 6023,
      "task_loss": 0.13124635815620422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976846246486426,
      "compression_loss": 0.0,
      "distillation_loss": 0.010580083355307579,
      "epoch": 5.72,
      "learning_rate": 4.0312048926907156e-06,
      "loss": 0.0099,
      "step": 6024,
      "task_loss": 0.0037852171808481216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976875164576563,
      "compression_loss": 0.0,
      "distillation_loss": 0.036560431122779846,
      "epoch": 5.72,
      "learning_rate": 4.025404867765103e-06,
      "loss": 0.0468,
      "step": 6025,
      "task_loss": 0.13939811289310455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797690405857833,
      "compression_loss": 0.0,
      "distillation_loss": 0.05065472796559334,
      "epoch": 5.72,
      "learning_rate": 4.019608652975712e-06,
      "loss": 0.0525,
      "step": 6026,
      "task_loss": 0.06941260397434235
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976932928501764,
      "compression_loss": 0.0,
      "distillation_loss": 0.033366162329912186,
      "epoch": 5.72,
      "learning_rate": 4.013816249375446e-06,
      "loss": 0.0437,
      "step": 6027,
      "task_loss": 0.13670209050178528
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976961774356901,
      "compression_loss": 0.0,
      "distillation_loss": 0.07585856318473816,
      "epoch": 5.72,
      "learning_rate": 4.008027658016536e-06,
      "loss": 0.0709,
      "step": 6028,
      "task_loss": 0.026351409032940865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7976990596153778,
      "compression_loss": 0.0,
      "distillation_loss": 0.15275973081588745,
      "epoch": 5.73,
      "learning_rate": 4.0022428799504936e-06,
      "loss": 0.1562,
      "step": 6029,
      "task_loss": 0.18706992268562317
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977019393902433,
      "compression_loss": 0.0,
      "distillation_loss": 0.014029610902071,
      "epoch": 5.73,
      "learning_rate": 3.996461916228159e-06,
      "loss": 0.02,
      "step": 6030,
      "task_loss": 0.07381336390972137
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79770481676129,
      "compression_loss": 0.0,
      "distillation_loss": 0.014353383332490921,
      "epoch": 5.73,
      "learning_rate": 3.990684767899677e-06,
      "loss": 0.0231,
      "step": 6031,
      "task_loss": 0.10191542655229568
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977076917295219,
      "compression_loss": 0.0,
      "distillation_loss": 0.12925012409687042,
      "epoch": 5.73,
      "learning_rate": 3.984911436014488e-06,
      "loss": 0.1234,
      "step": 6032,
      "task_loss": 0.07045982778072357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977105642959426,
      "compression_loss": 0.0,
      "distillation_loss": 0.021065477281808853,
      "epoch": 5.73,
      "learning_rate": 3.979141921621346e-06,
      "loss": 0.0263,
      "step": 6033,
      "task_loss": 0.07347995042800903
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977134344615556,
      "compression_loss": 0.0,
      "distillation_loss": 0.03473827242851257,
      "epoch": 5.73,
      "learning_rate": 3.9733762257683076e-06,
      "loss": 0.0325,
      "step": 6034,
      "task_loss": 0.012779684737324715
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977163022273648,
      "compression_loss": 0.0,
      "distillation_loss": 0.01884627155959606,
      "epoch": 5.73,
      "learning_rate": 3.967614349502749e-06,
      "loss": 0.0172,
      "step": 6035,
      "task_loss": 0.002278236672282219
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977191675943738,
      "compression_loss": 0.0,
      "distillation_loss": 0.021827183663845062,
      "epoch": 5.73,
      "learning_rate": 3.961856293871336e-06,
      "loss": 0.0203,
      "step": 6036,
      "task_loss": 0.0063978079706430435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977220305635863,
      "compression_loss": 0.0,
      "distillation_loss": 0.0746820718050003,
      "epoch": 5.73,
      "learning_rate": 3.956102059920036e-06,
      "loss": 0.0857,
      "step": 6037,
      "task_loss": 0.1851119101047516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977248911360059,
      "compression_loss": 0.0,
      "distillation_loss": 0.04180888831615448,
      "epoch": 5.73,
      "learning_rate": 3.9503516486941565e-06,
      "loss": 0.039,
      "step": 6038,
      "task_loss": 0.014088783413171768
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977277493126363,
      "compression_loss": 0.0,
      "distillation_loss": 0.02447015978395939,
      "epoch": 5.74,
      "learning_rate": 3.944605061238277e-06,
      "loss": 0.03,
      "step": 6039,
      "task_loss": 0.07963259518146515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977306050944813,
      "compression_loss": 0.0,
      "distillation_loss": 0.023615635931491852,
      "epoch": 5.74,
      "learning_rate": 3.938862298596291e-06,
      "loss": 0.0276,
      "step": 6040,
      "task_loss": 0.06358890980482101
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977334584825445,
      "compression_loss": 0.0,
      "distillation_loss": 0.03558419644832611,
      "epoch": 5.74,
      "learning_rate": 3.9331233618113974e-06,
      "loss": 0.0409,
      "step": 6041,
      "task_loss": 0.0888403058052063
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977363094778297,
      "compression_loss": 0.0,
      "distillation_loss": 0.016286537051200867,
      "epoch": 5.74,
      "learning_rate": 3.9273882519261066e-06,
      "loss": 0.0152,
      "step": 6042,
      "task_loss": 0.005049385130405426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977391580813403,
      "compression_loss": 0.0,
      "distillation_loss": 0.03099142014980316,
      "epoch": 5.74,
      "learning_rate": 3.921656969982229e-06,
      "loss": 0.04,
      "step": 6043,
      "task_loss": 0.12095896154642105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977420042940803,
      "compression_loss": 0.0,
      "distillation_loss": 0.09785278886556625,
      "epoch": 5.74,
      "learning_rate": 3.9159295170208695e-06,
      "loss": 0.1007,
      "step": 6044,
      "task_loss": 0.126520037651062
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797744848117053,
      "compression_loss": 0.0,
      "distillation_loss": 0.015856163576245308,
      "epoch": 5.74,
      "learning_rate": 3.910205894082461e-06,
      "loss": 0.0301,
      "step": 6045,
      "task_loss": 0.15833263099193573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977476895512625,
      "compression_loss": 0.0,
      "distillation_loss": 0.06277845799922943,
      "epoch": 5.74,
      "learning_rate": 3.904486102206717e-06,
      "loss": 0.0659,
      "step": 6046,
      "task_loss": 0.09369910508394241
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977505285977123,
      "compression_loss": 0.0,
      "distillation_loss": 0.05015091598033905,
      "epoch": 5.74,
      "learning_rate": 3.898770142432676e-06,
      "loss": 0.0622,
      "step": 6047,
      "task_loss": 0.1709066480398178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797753365257406,
      "compression_loss": 0.0,
      "distillation_loss": 0.02492024190723896,
      "epoch": 5.74,
      "learning_rate": 3.893058015798656e-06,
      "loss": 0.0323,
      "step": 6048,
      "task_loss": 0.09882645308971405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977561995313475,
      "compression_loss": 0.0,
      "distillation_loss": 0.017794858664274216,
      "epoch": 5.74,
      "learning_rate": 3.887349723342304e-06,
      "loss": 0.0305,
      "step": 6049,
      "task_loss": 0.14453651010990143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977590314205403,
      "compression_loss": 0.0,
      "distillation_loss": 0.023934083059430122,
      "epoch": 5.75,
      "learning_rate": 3.881645266100556e-06,
      "loss": 0.0297,
      "step": 6050,
      "task_loss": 0.08163676410913467
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977618609259881,
      "compression_loss": 0.0,
      "distillation_loss": 0.04733239859342575,
      "epoch": 5.75,
      "learning_rate": 3.875944645109647e-06,
      "loss": 0.0493,
      "step": 6051,
      "task_loss": 0.06739036738872528
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977646880486946,
      "compression_loss": 0.0,
      "distillation_loss": 0.07677607238292694,
      "epoch": 5.75,
      "learning_rate": 3.8702478614051355e-06,
      "loss": 0.0831,
      "step": 6052,
      "task_loss": 0.14044952392578125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977675127896635,
      "compression_loss": 0.0,
      "distillation_loss": 0.08341547846794128,
      "epoch": 5.75,
      "learning_rate": 3.864554916021859e-06,
      "loss": 0.08,
      "step": 6053,
      "task_loss": 0.04903567209839821
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977703351498985,
      "compression_loss": 0.0,
      "distillation_loss": 0.04993259161710739,
      "epoch": 5.75,
      "learning_rate": 3.8588658099939665e-06,
      "loss": 0.0565,
      "step": 6054,
      "task_loss": 0.11562397330999374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977731551304031,
      "compression_loss": 0.0,
      "distillation_loss": 0.019889146089553833,
      "epoch": 5.75,
      "learning_rate": 3.853180544354915e-06,
      "loss": 0.0268,
      "step": 6055,
      "task_loss": 0.0885215774178505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977759727321814,
      "compression_loss": 0.0,
      "distillation_loss": 0.028516318649053574,
      "epoch": 5.75,
      "learning_rate": 3.847499120137471e-06,
      "loss": 0.0281,
      "step": 6056,
      "task_loss": 0.024541892111301422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977787879562367,
      "compression_loss": 0.0,
      "distillation_loss": 0.01963622495532036,
      "epoch": 5.75,
      "learning_rate": 3.84182153837368e-06,
      "loss": 0.0326,
      "step": 6057,
      "task_loss": 0.1491500735282898
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977816008035727,
      "compression_loss": 0.0,
      "distillation_loss": 0.04049837589263916,
      "epoch": 5.75,
      "learning_rate": 3.836147800094903e-06,
      "loss": 0.0401,
      "step": 6058,
      "task_loss": 0.036487508565187454
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977844112751933,
      "compression_loss": 0.0,
      "distillation_loss": 0.059603024274110794,
      "epoch": 5.75,
      "learning_rate": 3.830477906331806e-06,
      "loss": 0.0588,
      "step": 6059,
      "task_loss": 0.05121609568595886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977872193721021,
      "compression_loss": 0.0,
      "distillation_loss": 0.03637555241584778,
      "epoch": 5.75,
      "learning_rate": 3.824811858114355e-06,
      "loss": 0.0334,
      "step": 6060,
      "task_loss": 0.006767544895410538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977900250953027,
      "compression_loss": 0.0,
      "distillation_loss": 0.023839876055717468,
      "epoch": 5.76,
      "learning_rate": 3.819149656471802e-06,
      "loss": 0.0227,
      "step": 6061,
      "task_loss": 0.012284144759178162
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977928284457988,
      "compression_loss": 0.0,
      "distillation_loss": 0.012317647226154804,
      "epoch": 5.76,
      "learning_rate": 3.8134913024327284e-06,
      "loss": 0.0171,
      "step": 6062,
      "task_loss": 0.0602097250521183
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977956294245941,
      "compression_loss": 0.0,
      "distillation_loss": 0.04070582613348961,
      "epoch": 5.76,
      "learning_rate": 3.8078367970249885e-06,
      "loss": 0.0551,
      "step": 6063,
      "task_loss": 0.18498341739177704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7977984280326924,
      "compression_loss": 0.0,
      "distillation_loss": 0.06427805125713348,
      "epoch": 5.76,
      "learning_rate": 3.8021861412757654e-06,
      "loss": 0.069,
      "step": 6064,
      "task_loss": 0.11124201118946075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978012242710972,
      "compression_loss": 0.0,
      "distillation_loss": 0.13212275505065918,
      "epoch": 5.76,
      "learning_rate": 3.796539336211513e-06,
      "loss": 0.1268,
      "step": 6065,
      "task_loss": 0.07930596172809601
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978040181408124,
      "compression_loss": 0.0,
      "distillation_loss": 0.0323861762881279,
      "epoch": 5.76,
      "learning_rate": 3.790896382858011e-06,
      "loss": 0.0336,
      "step": 6066,
      "task_loss": 0.04439293220639229
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978068096428413,
      "compression_loss": 0.0,
      "distillation_loss": 0.05183859169483185,
      "epoch": 5.76,
      "learning_rate": 3.785257282240326e-06,
      "loss": 0.0565,
      "step": 6067,
      "task_loss": 0.09798099845647812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797809598778188,
      "compression_loss": 0.0,
      "distillation_loss": 0.02488609589636326,
      "epoch": 5.76,
      "learning_rate": 3.779622035382821e-06,
      "loss": 0.0376,
      "step": 6068,
      "task_loss": 0.15173770487308502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978123855478559,
      "compression_loss": 0.0,
      "distillation_loss": 0.013482201844453812,
      "epoch": 5.76,
      "learning_rate": 3.7739906433091804e-06,
      "loss": 0.0253,
      "step": 6069,
      "task_loss": 0.1319684386253357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797815169952849,
      "compression_loss": 0.0,
      "distillation_loss": 0.013858940452337265,
      "epoch": 5.76,
      "learning_rate": 3.768363107042361e-06,
      "loss": 0.0128,
      "step": 6070,
      "task_loss": 0.0029547102749347687
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978179519941705,
      "compression_loss": 0.0,
      "distillation_loss": 0.030178038403391838,
      "epoch": 5.77,
      "learning_rate": 3.7627394276046337e-06,
      "loss": 0.0281,
      "step": 6071,
      "task_loss": 0.009721649810671806
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978207316728245,
      "compression_loss": 0.0,
      "distillation_loss": 0.036198072135448456,
      "epoch": 5.77,
      "learning_rate": 3.7571196060175672e-06,
      "loss": 0.0389,
      "step": 6072,
      "task_loss": 0.06324063241481781
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978235089898145,
      "compression_loss": 0.0,
      "distillation_loss": 0.03386306017637253,
      "epoch": 5.77,
      "learning_rate": 3.7515036433020343e-06,
      "loss": 0.0425,
      "step": 6073,
      "task_loss": 0.12017367035150528
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978262839461442,
      "compression_loss": 0.0,
      "distillation_loss": 0.03635266050696373,
      "epoch": 5.77,
      "learning_rate": 3.7458915404782024e-06,
      "loss": 0.0361,
      "step": 6074,
      "task_loss": 0.034307606518268585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978290565428173,
      "compression_loss": 0.0,
      "distillation_loss": 0.04068465903401375,
      "epoch": 5.77,
      "learning_rate": 3.740283298565525e-06,
      "loss": 0.0485,
      "step": 6075,
      "task_loss": 0.11905878782272339
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978318267808375,
      "compression_loss": 0.0,
      "distillation_loss": 0.10536225140094757,
      "epoch": 5.77,
      "learning_rate": 3.734678918582782e-06,
      "loss": 0.1052,
      "step": 6076,
      "task_loss": 0.10353413224220276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978345946612085,
      "compression_loss": 0.0,
      "distillation_loss": 0.05913068726658821,
      "epoch": 5.77,
      "learning_rate": 3.729078401548028e-06,
      "loss": 0.0667,
      "step": 6077,
      "task_loss": 0.1346680372953415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978373601849339,
      "compression_loss": 0.0,
      "distillation_loss": 0.029158534482121468,
      "epoch": 5.77,
      "learning_rate": 3.723481748478619e-06,
      "loss": 0.038,
      "step": 6078,
      "task_loss": 0.11800020188093185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978401233530175,
      "compression_loss": 0.0,
      "distillation_loss": 0.023293327540159225,
      "epoch": 5.77,
      "learning_rate": 3.717888960391222e-06,
      "loss": 0.0218,
      "step": 6079,
      "task_loss": 0.008854774758219719
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978428841664628,
      "compression_loss": 0.0,
      "distillation_loss": 0.06899020075798035,
      "epoch": 5.77,
      "learning_rate": 3.7123000383017944e-06,
      "loss": 0.063,
      "step": 6080,
      "task_loss": 0.009547203779220581
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978456426262737,
      "compression_loss": 0.0,
      "distillation_loss": 0.027045968919992447,
      "epoch": 5.77,
      "learning_rate": 3.7067149832255787e-06,
      "loss": 0.0329,
      "step": 6081,
      "task_loss": 0.08595190942287445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978483987334537,
      "compression_loss": 0.0,
      "distillation_loss": 0.18197451531887054,
      "epoch": 5.78,
      "learning_rate": 3.7011337961771394e-06,
      "loss": 0.1829,
      "step": 6082,
      "task_loss": 0.19158345460891724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978511524890065,
      "compression_loss": 0.0,
      "distillation_loss": 0.104429692029953,
      "epoch": 5.78,
      "learning_rate": 3.6955564781703257e-06,
      "loss": 0.1017,
      "step": 6083,
      "task_loss": 0.07724487781524658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797853903893936,
      "compression_loss": 0.0,
      "distillation_loss": 0.05032842978835106,
      "epoch": 5.78,
      "learning_rate": 3.6899830302182832e-06,
      "loss": 0.0553,
      "step": 6084,
      "task_loss": 0.1004081666469574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978566529492456,
      "compression_loss": 0.0,
      "distillation_loss": 0.02652582712471485,
      "epoch": 5.78,
      "learning_rate": 3.6844134533334437e-06,
      "loss": 0.0353,
      "step": 6085,
      "task_loss": 0.11424309015274048
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978593996559391,
      "compression_loss": 0.0,
      "distillation_loss": 0.03817993402481079,
      "epoch": 5.78,
      "learning_rate": 3.6788477485275667e-06,
      "loss": 0.0446,
      "step": 6086,
      "task_loss": 0.1024344339966774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978621440150202,
      "compression_loss": 0.0,
      "distillation_loss": 0.04021666944026947,
      "epoch": 5.78,
      "learning_rate": 3.673285916811678e-06,
      "loss": 0.0425,
      "step": 6087,
      "task_loss": 0.06331319361925125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978648860274926,
      "compression_loss": 0.0,
      "distillation_loss": 0.021305495873093605,
      "epoch": 5.78,
      "learning_rate": 3.66772795919611e-06,
      "loss": 0.0294,
      "step": 6088,
      "task_loss": 0.10244767367839813
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79786762569436,
      "compression_loss": 0.0,
      "distillation_loss": 0.05766810476779938,
      "epoch": 5.78,
      "learning_rate": 3.6621738766904944e-06,
      "loss": 0.0829,
      "step": 6089,
      "task_loss": 0.3096805214881897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797870363016626,
      "compression_loss": 0.0,
      "distillation_loss": 0.06154272332787514,
      "epoch": 5.78,
      "learning_rate": 3.6566236703037633e-06,
      "loss": 0.0717,
      "step": 6090,
      "task_loss": 0.1632338911294937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978730979952942,
      "compression_loss": 0.0,
      "distillation_loss": 0.11370549350976944,
      "epoch": 5.78,
      "learning_rate": 3.6510773410441316e-06,
      "loss": 0.1306,
      "step": 6091,
      "task_loss": 0.2824677526950836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978758306313685,
      "compression_loss": 0.0,
      "distillation_loss": 0.030081721022725105,
      "epoch": 5.79,
      "learning_rate": 3.6455348899191137e-06,
      "loss": 0.0389,
      "step": 6092,
      "task_loss": 0.11794061213731766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978785609258524,
      "compression_loss": 0.0,
      "distillation_loss": 0.01757514476776123,
      "epoch": 5.79,
      "learning_rate": 3.6399963179355347e-06,
      "loss": 0.0166,
      "step": 6093,
      "task_loss": 0.007390303537249565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978812888797497,
      "compression_loss": 0.0,
      "distillation_loss": 0.019966866821050644,
      "epoch": 5.79,
      "learning_rate": 3.6344616260994946e-06,
      "loss": 0.0241,
      "step": 6094,
      "task_loss": 0.06160247325897217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978840144940641,
      "compression_loss": 0.0,
      "distillation_loss": 0.03950147703289986,
      "epoch": 5.79,
      "learning_rate": 3.6289308154163927e-06,
      "loss": 0.0375,
      "step": 6095,
      "task_loss": 0.019919494166970253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978867377697991,
      "compression_loss": 0.0,
      "distillation_loss": 0.02330363541841507,
      "epoch": 5.79,
      "learning_rate": 3.6234038868909357e-06,
      "loss": 0.0384,
      "step": 6096,
      "task_loss": 0.17430183291435242
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978894587079586,
      "compression_loss": 0.0,
      "distillation_loss": 0.020831573754549026,
      "epoch": 5.79,
      "learning_rate": 3.6178808415271158e-06,
      "loss": 0.0274,
      "step": 6097,
      "task_loss": 0.08686920255422592
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978921773095462,
      "compression_loss": 0.0,
      "distillation_loss": 0.06831928342580795,
      "epoch": 5.79,
      "learning_rate": 3.6123616803282132e-06,
      "loss": 0.0654,
      "step": 6098,
      "task_loss": 0.03933015465736389
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978948935755655,
      "compression_loss": 0.0,
      "distillation_loss": 0.033079955726861954,
      "epoch": 5.79,
      "learning_rate": 3.6068464042968123e-06,
      "loss": 0.0432,
      "step": 6099,
      "task_loss": 0.1338474154472351
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7978976075070203,
      "compression_loss": 0.0,
      "distillation_loss": 0.01981043629348278,
      "epoch": 5.79,
      "learning_rate": 3.6013350144348005e-06,
      "loss": 0.0185,
      "step": 6100,
      "task_loss": 0.00669914111495018
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979003191049142,
      "compression_loss": 0.0,
      "distillation_loss": 0.05477694794535637,
      "epoch": 5.79,
      "learning_rate": 3.595827511743341e-06,
      "loss": 0.0621,
      "step": 6101,
      "task_loss": 0.12840932607650757
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797903028370251,
      "compression_loss": 0.0,
      "distillation_loss": 0.04909781366586685,
      "epoch": 5.79,
      "learning_rate": 3.5903238972228916e-06,
      "loss": 0.0456,
      "step": 6102,
      "task_loss": 0.014244996011257172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979057353040342,
      "compression_loss": 0.0,
      "distillation_loss": 0.015033694915473461,
      "epoch": 5.8,
      "learning_rate": 3.584824171873219e-06,
      "loss": 0.014,
      "step": 6103,
      "task_loss": 0.004645902663469315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979084399072676,
      "compression_loss": 0.0,
      "distillation_loss": 0.03330419957637787,
      "epoch": 5.8,
      "learning_rate": 3.579328336693377e-06,
      "loss": 0.0474,
      "step": 6104,
      "task_loss": 0.17442171275615692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979111421809548,
      "compression_loss": 0.0,
      "distillation_loss": 0.03729955852031708,
      "epoch": 5.8,
      "learning_rate": 3.5738363926816954e-06,
      "loss": 0.0373,
      "step": 6105,
      "task_loss": 0.03731003403663635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979138421260996,
      "compression_loss": 0.0,
      "distillation_loss": 0.028155038133263588,
      "epoch": 5.8,
      "learning_rate": 3.5683483408358307e-06,
      "loss": 0.0345,
      "step": 6106,
      "task_loss": 0.09175101667642593
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979165397437057,
      "compression_loss": 0.0,
      "distillation_loss": 0.02423456311225891,
      "epoch": 5.8,
      "learning_rate": 3.5628641821527002e-06,
      "loss": 0.0265,
      "step": 6107,
      "task_loss": 0.04681221395730972
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979192350347767,
      "compression_loss": 0.0,
      "distillation_loss": 0.025493551045656204,
      "epoch": 5.8,
      "learning_rate": 3.55738391762854e-06,
      "loss": 0.0236,
      "step": 6108,
      "task_loss": 0.00649440661072731
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979219280003162,
      "compression_loss": 0.0,
      "distillation_loss": 0.0265052393078804,
      "epoch": 5.8,
      "learning_rate": 3.551907548258854e-06,
      "loss": 0.0333,
      "step": 6109,
      "task_loss": 0.09423819929361343
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797924618641328,
      "compression_loss": 0.0,
      "distillation_loss": 0.019546575844287872,
      "epoch": 5.8,
      "learning_rate": 3.5464350750384594e-06,
      "loss": 0.0277,
      "step": 6110,
      "task_loss": 0.10154897719621658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979273069588158,
      "compression_loss": 0.0,
      "distillation_loss": 0.06870542466640472,
      "epoch": 5.8,
      "learning_rate": 3.5409664989614593e-06,
      "loss": 0.0715,
      "step": 6111,
      "task_loss": 0.09712858498096466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979299929537832,
      "compression_loss": 0.0,
      "distillation_loss": 0.019337208941578865,
      "epoch": 5.8,
      "learning_rate": 3.5355018210212394e-06,
      "loss": 0.0179,
      "step": 6112,
      "task_loss": 0.004613950848579407
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979326766272339,
      "compression_loss": 0.0,
      "distillation_loss": 0.020329779013991356,
      "epoch": 5.81,
      "learning_rate": 3.530041042210483e-06,
      "loss": 0.0191,
      "step": 6113,
      "task_loss": 0.008153880015015602
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979353579801717,
      "compression_loss": 0.0,
      "distillation_loss": 0.06501001864671707,
      "epoch": 5.81,
      "learning_rate": 3.524584163521177e-06,
      "loss": 0.0633,
      "step": 6114,
      "task_loss": 0.048086702823638916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979380370136001,
      "compression_loss": 0.0,
      "distillation_loss": 0.026702333241701126,
      "epoch": 5.81,
      "learning_rate": 3.5191311859445796e-06,
      "loss": 0.0326,
      "step": 6115,
      "task_loss": 0.08566763997077942
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979407137285228,
      "compression_loss": 0.0,
      "distillation_loss": 0.07300684601068497,
      "epoch": 5.81,
      "learning_rate": 3.5136821104712557e-06,
      "loss": 0.0717,
      "step": 6116,
      "task_loss": 0.059550218284130096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979433881259437,
      "compression_loss": 0.0,
      "distillation_loss": 0.023176930844783783,
      "epoch": 5.81,
      "learning_rate": 3.5082369380910612e-06,
      "loss": 0.0228,
      "step": 6117,
      "task_loss": 0.01990087516605854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979460602068662,
      "compression_loss": 0.0,
      "distillation_loss": 0.01764390803873539,
      "epoch": 5.81,
      "learning_rate": 3.5027956697931324e-06,
      "loss": 0.0205,
      "step": 6118,
      "task_loss": 0.04635511338710785
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979487299722942,
      "compression_loss": 0.0,
      "distillation_loss": 0.049246713519096375,
      "epoch": 5.81,
      "learning_rate": 3.497358306565901e-06,
      "loss": 0.055,
      "step": 6119,
      "task_loss": 0.1070106029510498
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979513974232312,
      "compression_loss": 0.0,
      "distillation_loss": 0.024939920753240585,
      "epoch": 5.81,
      "learning_rate": 3.491924849397085e-06,
      "loss": 0.0399,
      "step": 6120,
      "task_loss": 0.1745125949382782
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797954062560681,
      "compression_loss": 0.0,
      "distillation_loss": 0.13209006190299988,
      "epoch": 5.81,
      "learning_rate": 3.486495299273712e-06,
      "loss": 0.1453,
      "step": 6121,
      "task_loss": 0.2643681466579437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979567253856473,
      "compression_loss": 0.0,
      "distillation_loss": 0.02114560827612877,
      "epoch": 5.81,
      "learning_rate": 3.481069657182076e-06,
      "loss": 0.0195,
      "step": 6122,
      "task_loss": 0.004614364355802536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979593858991337,
      "compression_loss": 0.0,
      "distillation_loss": 0.05115702748298645,
      "epoch": 5.81,
      "learning_rate": 3.4756479241077667e-06,
      "loss": 0.0508,
      "step": 6123,
      "task_loss": 0.04790136218070984
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797962044102144,
      "compression_loss": 0.0,
      "distillation_loss": 0.015820186585187912,
      "epoch": 5.82,
      "learning_rate": 3.4702301010356762e-06,
      "loss": 0.0145,
      "step": 6124,
      "task_loss": 0.002711120992898941
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979646999956818,
      "compression_loss": 0.0,
      "distillation_loss": 0.049572013318538666,
      "epoch": 5.82,
      "learning_rate": 3.464816188949982e-06,
      "loss": 0.0551,
      "step": 6125,
      "task_loss": 0.10523169487714767
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979673535807508,
      "compression_loss": 0.0,
      "distillation_loss": 0.020622238516807556,
      "epoch": 5.82,
      "learning_rate": 3.459406188834141e-06,
      "loss": 0.0318,
      "step": 6126,
      "task_loss": 0.13285362720489502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979700048583546,
      "compression_loss": 0.0,
      "distillation_loss": 0.027139317244291306,
      "epoch": 5.82,
      "learning_rate": 3.454000101670901e-06,
      "loss": 0.0251,
      "step": 6127,
      "task_loss": 0.006299933418631554
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979726538294969,
      "compression_loss": 0.0,
      "distillation_loss": 0.008850215002894402,
      "epoch": 5.82,
      "learning_rate": 3.4485979284423155e-06,
      "loss": 0.0129,
      "step": 6128,
      "task_loss": 0.04908328130841255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979753004951815,
      "compression_loss": 0.0,
      "distillation_loss": 0.031228946521878242,
      "epoch": 5.82,
      "learning_rate": 3.443199670129707e-06,
      "loss": 0.0441,
      "step": 6129,
      "task_loss": 0.15944904088974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979779448564119,
      "compression_loss": 0.0,
      "distillation_loss": 0.13153165578842163,
      "epoch": 5.82,
      "learning_rate": 3.4378053277136946e-06,
      "loss": 0.1276,
      "step": 6130,
      "task_loss": 0.09214229881763458
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.797980586914192,
      "compression_loss": 0.0,
      "distillation_loss": 0.04689065366983414,
      "epoch": 5.82,
      "learning_rate": 3.4324149021741915e-06,
      "loss": 0.0474,
      "step": 6131,
      "task_loss": 0.05218418687582016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979832266695255,
      "compression_loss": 0.0,
      "distillation_loss": 0.0955977588891983,
      "epoch": 5.82,
      "learning_rate": 3.4270283944903946e-06,
      "loss": 0.087,
      "step": 6132,
      "task_loss": 0.009947611019015312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979858641234158,
      "compression_loss": 0.0,
      "distillation_loss": 0.11793653666973114,
      "epoch": 5.82,
      "learning_rate": 3.4216458056407775e-06,
      "loss": 0.1168,
      "step": 6133,
      "task_loss": 0.10687369108200073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979884992768668,
      "compression_loss": 0.0,
      "distillation_loss": 0.016993030905723572,
      "epoch": 5.83,
      "learning_rate": 3.416267136603124e-06,
      "loss": 0.016,
      "step": 6134,
      "task_loss": 0.006749266758561134
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979911321308821,
      "compression_loss": 0.0,
      "distillation_loss": 0.03799954056739807,
      "epoch": 5.83,
      "learning_rate": 3.4108923883545004e-06,
      "loss": 0.0412,
      "step": 6135,
      "task_loss": 0.0700988918542862
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979937626864654,
      "compression_loss": 0.0,
      "distillation_loss": 0.02764744684100151,
      "epoch": 5.83,
      "learning_rate": 3.405521561871247e-06,
      "loss": 0.0338,
      "step": 6136,
      "task_loss": 0.08966173231601715
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979963909446204,
      "compression_loss": 0.0,
      "distillation_loss": 0.017384584993124008,
      "epoch": 5.83,
      "learning_rate": 3.400154658128998e-06,
      "loss": 0.0161,
      "step": 6137,
      "task_loss": 0.004670457914471626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7979990169063508,
      "compression_loss": 0.0,
      "distillation_loss": 0.029262276366353035,
      "epoch": 5.83,
      "learning_rate": 3.3947916781026877e-06,
      "loss": 0.0339,
      "step": 6138,
      "task_loss": 0.07558421790599823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980016405726602,
      "compression_loss": 0.0,
      "distillation_loss": 0.02195669896900654,
      "epoch": 5.83,
      "learning_rate": 3.3894326227665196e-06,
      "loss": 0.0203,
      "step": 6139,
      "task_loss": 0.005175255239009857
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980042619445524,
      "compression_loss": 0.0,
      "distillation_loss": 0.017028218135237694,
      "epoch": 5.83,
      "learning_rate": 3.384077493093987e-06,
      "loss": 0.0157,
      "step": 6140,
      "task_loss": 0.003494156524538994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798006881023031,
      "compression_loss": 0.0,
      "distillation_loss": 0.01528292428702116,
      "epoch": 5.83,
      "learning_rate": 3.378726290057882e-06,
      "loss": 0.0142,
      "step": 6141,
      "task_loss": 0.004559960216283798
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980094978090997,
      "compression_loss": 0.0,
      "distillation_loss": 0.014390531927347183,
      "epoch": 5.83,
      "learning_rate": 3.373379014630279e-06,
      "loss": 0.02,
      "step": 6142,
      "task_loss": 0.07083210349082947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980121123037621,
      "compression_loss": 0.0,
      "distillation_loss": 0.0584423765540123,
      "epoch": 5.83,
      "learning_rate": 3.368035667782535e-06,
      "loss": 0.0641,
      "step": 6143,
      "task_loss": 0.11510256677865982
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798014724508022,
      "compression_loss": 0.0,
      "distillation_loss": 0.035305943340063095,
      "epoch": 5.83,
      "learning_rate": 3.3626962504852865e-06,
      "loss": 0.0345,
      "step": 6144,
      "task_loss": 0.026764407753944397
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980173344228831,
      "compression_loss": 0.0,
      "distillation_loss": 0.026942508295178413,
      "epoch": 5.84,
      "learning_rate": 3.357360763708478e-06,
      "loss": 0.0398,
      "step": 6145,
      "task_loss": 0.15590748190879822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798019942049349,
      "compression_loss": 0.0,
      "distillation_loss": 0.10806174576282501,
      "epoch": 5.84,
      "learning_rate": 3.3520292084213145e-06,
      "loss": 0.1066,
      "step": 6146,
      "task_loss": 0.09339408576488495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980225473884235,
      "compression_loss": 0.0,
      "distillation_loss": 0.03551897406578064,
      "epoch": 5.84,
      "learning_rate": 3.3467015855923016e-06,
      "loss": 0.0324,
      "step": 6147,
      "task_loss": 0.004185806959867477
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79802515044111,
      "compression_loss": 0.0,
      "distillation_loss": 0.18967923521995544,
      "epoch": 5.84,
      "learning_rate": 3.3413778961892324e-06,
      "loss": 0.1758,
      "step": 6148,
      "task_loss": 0.05130823701620102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980277512084125,
      "compression_loss": 0.0,
      "distillation_loss": 0.0280881579965353,
      "epoch": 5.84,
      "learning_rate": 3.33605814117918e-06,
      "loss": 0.0319,
      "step": 6149,
      "task_loss": 0.06627509742975235
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980303496913346,
      "compression_loss": 0.0,
      "distillation_loss": 0.014908300712704659,
      "epoch": 5.84,
      "learning_rate": 3.3307423215284933e-06,
      "loss": 0.0137,
      "step": 6150,
      "task_loss": 0.002931041643023491
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980329458908798,
      "compression_loss": 0.0,
      "distillation_loss": 0.04687492176890373,
      "epoch": 5.84,
      "learning_rate": 3.325430438202823e-06,
      "loss": 0.0432,
      "step": 6151,
      "task_loss": 0.009853720664978027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798035539808052,
      "compression_loss": 0.0,
      "distillation_loss": 0.08521527796983719,
      "epoch": 5.84,
      "learning_rate": 3.320122492167108e-06,
      "loss": 0.0926,
      "step": 6152,
      "task_loss": 0.15886865556240082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980381314438548,
      "compression_loss": 0.0,
      "distillation_loss": 0.023487182334065437,
      "epoch": 5.84,
      "learning_rate": 3.3148184843855513e-06,
      "loss": 0.0218,
      "step": 6153,
      "task_loss": 0.0069718193262815475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980407207992919,
      "compression_loss": 0.0,
      "distillation_loss": 0.008888408541679382,
      "epoch": 5.84,
      "learning_rate": 3.3095184158216497e-06,
      "loss": 0.0085,
      "step": 6154,
      "task_loss": 0.005421014502644539
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980433078753669,
      "compression_loss": 0.0,
      "distillation_loss": 0.014762504026293755,
      "epoch": 5.85,
      "learning_rate": 3.304222287438194e-06,
      "loss": 0.0202,
      "step": 6155,
      "task_loss": 0.06863877922296524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980458926730837,
      "compression_loss": 0.0,
      "distillation_loss": 0.09593481570482254,
      "epoch": 5.85,
      "learning_rate": 3.2989301001972505e-06,
      "loss": 0.1038,
      "step": 6156,
      "task_loss": 0.17508172988891602
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980484751934457,
      "compression_loss": 0.0,
      "distillation_loss": 0.04558049887418747,
      "epoch": 5.85,
      "learning_rate": 3.293641855060162e-06,
      "loss": 0.0424,
      "step": 6157,
      "task_loss": 0.013653889298439026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980510554374567,
      "compression_loss": 0.0,
      "distillation_loss": 0.021937305107712746,
      "epoch": 5.85,
      "learning_rate": 3.2883575529875726e-06,
      "loss": 0.0212,
      "step": 6158,
      "task_loss": 0.01466263085603714
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980536334061205,
      "compression_loss": 0.0,
      "distillation_loss": 0.02071615308523178,
      "epoch": 5.85,
      "learning_rate": 3.2830771949393934e-06,
      "loss": 0.0194,
      "step": 6159,
      "task_loss": 0.007383199408650398
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980562091004405,
      "compression_loss": 0.0,
      "distillation_loss": 0.02836211957037449,
      "epoch": 5.85,
      "learning_rate": 3.27780078187484e-06,
      "loss": 0.0417,
      "step": 6160,
      "task_loss": 0.16145291924476624
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980587825214207,
      "compression_loss": 0.0,
      "distillation_loss": 0.06770492345094681,
      "epoch": 5.85,
      "learning_rate": 3.272528314752382e-06,
      "loss": 0.0737,
      "step": 6161,
      "task_loss": 0.12756067514419556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980613536700646,
      "compression_loss": 0.0,
      "distillation_loss": 0.041868340224027634,
      "epoch": 5.85,
      "learning_rate": 3.2672597945298006e-06,
      "loss": 0.0427,
      "step": 6162,
      "task_loss": 0.04990936070680618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980639225473759,
      "compression_loss": 0.0,
      "distillation_loss": 0.012833887711167336,
      "epoch": 5.85,
      "learning_rate": 3.2619952221641436e-06,
      "loss": 0.0204,
      "step": 6163,
      "task_loss": 0.08840099722146988
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980664891543584,
      "compression_loss": 0.0,
      "distillation_loss": 0.07975931465625763,
      "epoch": 5.85,
      "learning_rate": 3.2567345986117437e-06,
      "loss": 0.0906,
      "step": 6164,
      "task_loss": 0.188090518116951
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980690534920156,
      "compression_loss": 0.0,
      "distillation_loss": 0.04811672866344452,
      "epoch": 5.85,
      "learning_rate": 3.251477924828222e-06,
      "loss": 0.0546,
      "step": 6165,
      "task_loss": 0.11324407160282135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980716155613513,
      "compression_loss": 0.0,
      "distillation_loss": 0.023004433140158653,
      "epoch": 5.86,
      "learning_rate": 3.2462252017684797e-06,
      "loss": 0.0273,
      "step": 6166,
      "task_loss": 0.06551149487495422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980741753633692,
      "compression_loss": 0.0,
      "distillation_loss": 0.039101071655750275,
      "epoch": 5.86,
      "learning_rate": 3.240976430386691e-06,
      "loss": 0.0382,
      "step": 6167,
      "task_loss": 0.030401092022657394
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980767328990729,
      "compression_loss": 0.0,
      "distillation_loss": 0.03869537636637688,
      "epoch": 5.86,
      "learning_rate": 3.2357316116363278e-06,
      "loss": 0.0375,
      "step": 6168,
      "task_loss": 0.026733241975307465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798079288169466,
      "compression_loss": 0.0,
      "distillation_loss": 0.016314208507537842,
      "epoch": 5.86,
      "learning_rate": 3.2304907464701384e-06,
      "loss": 0.0184,
      "step": 6169,
      "task_loss": 0.037635624408721924
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980818411755525,
      "compression_loss": 0.0,
      "distillation_loss": 0.15428408980369568,
      "epoch": 5.86,
      "learning_rate": 3.225253835840147e-06,
      "loss": 0.1555,
      "step": 6170,
      "task_loss": 0.16609327495098114
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980843919183357,
      "compression_loss": 0.0,
      "distillation_loss": 0.024836333468556404,
      "epoch": 5.86,
      "learning_rate": 3.2200208806976634e-06,
      "loss": 0.0298,
      "step": 6171,
      "task_loss": 0.07457348704338074
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980869403988196,
      "compression_loss": 0.0,
      "distillation_loss": 0.025460699573159218,
      "epoch": 5.86,
      "learning_rate": 3.2147918819932814e-06,
      "loss": 0.0235,
      "step": 6172,
      "task_loss": 0.00573544017970562
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980894866180076,
      "compression_loss": 0.0,
      "distillation_loss": 0.020409489050507545,
      "epoch": 5.86,
      "learning_rate": 3.209566840676875e-06,
      "loss": 0.0189,
      "step": 6173,
      "task_loss": 0.00523163378238678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980920305769037,
      "compression_loss": 0.0,
      "distillation_loss": 0.04639172554016113,
      "epoch": 5.86,
      "learning_rate": 3.204345757697591e-06,
      "loss": 0.0423,
      "step": 6174,
      "task_loss": 0.005238974466919899
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980945722765113,
      "compression_loss": 0.0,
      "distillation_loss": 0.07041463255882263,
      "epoch": 5.86,
      "learning_rate": 3.199128634003873e-06,
      "loss": 0.0673,
      "step": 6175,
      "task_loss": 0.03931209072470665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980971117178342,
      "compression_loss": 0.0,
      "distillation_loss": 0.10844946652650833,
      "epoch": 5.87,
      "learning_rate": 3.1939154705434267e-06,
      "loss": 0.1306,
      "step": 6176,
      "task_loss": 0.32991883158683777
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7980996489018761,
      "compression_loss": 0.0,
      "distillation_loss": 0.0225403793156147,
      "epoch": 5.87,
      "learning_rate": 3.188706268263261e-06,
      "loss": 0.0212,
      "step": 6177,
      "task_loss": 0.009059395641088486
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981021838296406,
      "compression_loss": 0.0,
      "distillation_loss": 0.0714748352766037,
      "epoch": 5.87,
      "learning_rate": 3.183501028109642e-06,
      "loss": 0.0773,
      "step": 6178,
      "task_loss": 0.1296701580286026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981047165021314,
      "compression_loss": 0.0,
      "distillation_loss": 0.043413858860731125,
      "epoch": 5.87,
      "learning_rate": 3.1782997510281352e-06,
      "loss": 0.0535,
      "step": 6179,
      "task_loss": 0.14417198300361633
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981072469203524,
      "compression_loss": 0.0,
      "distillation_loss": 0.044636182487010956,
      "epoch": 5.87,
      "learning_rate": 3.173102437963571e-06,
      "loss": 0.0413,
      "step": 6180,
      "task_loss": 0.011053688824176788
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798109775085307,
      "compression_loss": 0.0,
      "distillation_loss": 0.04771566390991211,
      "epoch": 5.87,
      "learning_rate": 3.1679090898600663e-06,
      "loss": 0.0465,
      "step": 6181,
      "task_loss": 0.035778842866420746
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798112300997999,
      "compression_loss": 0.0,
      "distillation_loss": 0.14893057942390442,
      "epoch": 5.87,
      "learning_rate": 3.1627197076610243e-06,
      "loss": 0.149,
      "step": 6182,
      "task_loss": 0.14932629466056824
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798114824659432,
      "compression_loss": 0.0,
      "distillation_loss": 0.04840100184082985,
      "epoch": 5.87,
      "learning_rate": 3.157534292309114e-06,
      "loss": 0.0514,
      "step": 6183,
      "task_loss": 0.07838822156190872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981173460706098,
      "compression_loss": 0.0,
      "distillation_loss": 0.04644680395722389,
      "epoch": 5.87,
      "learning_rate": 3.1523528447462925e-06,
      "loss": 0.0423,
      "step": 6184,
      "task_loss": 0.004648592323064804
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798119865232536,
      "compression_loss": 0.0,
      "distillation_loss": 0.12863226234912872,
      "epoch": 5.87,
      "learning_rate": 3.147175365913793e-06,
      "loss": 0.1241,
      "step": 6185,
      "task_loss": 0.08361361920833588
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981223821462143,
      "compression_loss": 0.0,
      "distillation_loss": 0.10540250688791275,
      "epoch": 5.87,
      "learning_rate": 3.1420018567521412e-06,
      "loss": 0.1011,
      "step": 6186,
      "task_loss": 0.06201855093240738
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981248968126484,
      "compression_loss": 0.0,
      "distillation_loss": 0.027073774486780167,
      "epoch": 5.88,
      "learning_rate": 3.136832318201119e-06,
      "loss": 0.0254,
      "step": 6187,
      "task_loss": 0.010364262387156487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981274092328421,
      "compression_loss": 0.0,
      "distillation_loss": 0.09114031493663788,
      "epoch": 5.88,
      "learning_rate": 3.1316667511997967e-06,
      "loss": 0.0891,
      "step": 6188,
      "task_loss": 0.07045449316501617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981299194077989,
      "compression_loss": 0.0,
      "distillation_loss": 0.017900224775075912,
      "epoch": 5.88,
      "learning_rate": 3.126505156686532e-06,
      "loss": 0.0242,
      "step": 6189,
      "task_loss": 0.08130859583616257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981324273385224,
      "compression_loss": 0.0,
      "distillation_loss": 0.011481826193630695,
      "epoch": 5.88,
      "learning_rate": 3.121347535598948e-06,
      "loss": 0.0107,
      "step": 6190,
      "task_loss": 0.003604454919695854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981349330260166,
      "compression_loss": 0.0,
      "distillation_loss": 0.02443830482661724,
      "epoch": 5.88,
      "learning_rate": 3.11619388887395e-06,
      "loss": 0.0247,
      "step": 6191,
      "task_loss": 0.026645643636584282
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981374364712849,
      "compression_loss": 0.0,
      "distillation_loss": 0.049847427755594254,
      "epoch": 5.88,
      "learning_rate": 3.111044217447731e-06,
      "loss": 0.0453,
      "step": 6192,
      "task_loss": 0.004831980913877487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981399376753311,
      "compression_loss": 0.0,
      "distillation_loss": 0.027359748259186745,
      "epoch": 5.88,
      "learning_rate": 3.105898522255743e-06,
      "loss": 0.0262,
      "step": 6193,
      "task_loss": 0.015636542811989784
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981424366391588,
      "compression_loss": 0.0,
      "distillation_loss": 0.02219046652317047,
      "epoch": 5.88,
      "learning_rate": 3.100756804232735e-06,
      "loss": 0.0209,
      "step": 6194,
      "task_loss": 0.008930448442697525
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981449333637719,
      "compression_loss": 0.0,
      "distillation_loss": 0.01434510201215744,
      "epoch": 5.88,
      "learning_rate": 3.095619064312719e-06,
      "loss": 0.0132,
      "step": 6195,
      "task_loss": 0.0025472547858953476
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981474278501739,
      "compression_loss": 0.0,
      "distillation_loss": 0.014208871871232986,
      "epoch": 5.88,
      "learning_rate": 3.0904853034289943e-06,
      "loss": 0.0166,
      "step": 6196,
      "task_loss": 0.03827598690986633
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981499200993685,
      "compression_loss": 0.0,
      "distillation_loss": 0.028200728818774223,
      "epoch": 5.89,
      "learning_rate": 3.085355522514136e-06,
      "loss": 0.0336,
      "step": 6197,
      "task_loss": 0.08197523653507233
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981524101123594,
      "compression_loss": 0.0,
      "distillation_loss": 0.04185293987393379,
      "epoch": 5.89,
      "learning_rate": 3.0802297224999805e-06,
      "loss": 0.0391,
      "step": 6198,
      "task_loss": 0.014529986307024956
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981548978901502,
      "compression_loss": 0.0,
      "distillation_loss": 0.02289126068353653,
      "epoch": 5.89,
      "learning_rate": 3.0751079043176673e-06,
      "loss": 0.0286,
      "step": 6199,
      "task_loss": 0.08047693967819214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981573834337446,
      "compression_loss": 0.0,
      "distillation_loss": 0.014290563762187958,
      "epoch": 5.89,
      "learning_rate": 3.0699900688975964e-06,
      "loss": 0.0134,
      "step": 6200,
      "task_loss": 0.005466701462864876
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981598667441465,
      "compression_loss": 0.0,
      "distillation_loss": 0.11609383672475815,
      "epoch": 5.89,
      "learning_rate": 3.064876217169446e-06,
      "loss": 0.1114,
      "step": 6201,
      "task_loss": 0.06948763132095337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981623478223594,
      "compression_loss": 0.0,
      "distillation_loss": 0.05835815891623497,
      "epoch": 5.89,
      "learning_rate": 3.05976635006216e-06,
      "loss": 0.0661,
      "step": 6202,
      "task_loss": 0.13556569814682007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798164826669387,
      "compression_loss": 0.0,
      "distillation_loss": 0.13488264381885529,
      "epoch": 5.89,
      "learning_rate": 3.0546604685039947e-06,
      "loss": 0.1469,
      "step": 6203,
      "task_loss": 0.25527501106262207
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798167303286233,
      "compression_loss": 0.0,
      "distillation_loss": 0.052829310297966,
      "epoch": 5.89,
      "learning_rate": 3.0495585734224424e-06,
      "loss": 0.059,
      "step": 6204,
      "task_loss": 0.1148587167263031
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981697776739011,
      "compression_loss": 0.0,
      "distillation_loss": 0.01467627938836813,
      "epoch": 5.89,
      "learning_rate": 3.044460665744284e-06,
      "loss": 0.0242,
      "step": 6205,
      "task_loss": 0.11007630825042725
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981722498333949,
      "compression_loss": 0.0,
      "distillation_loss": 0.017747841775417328,
      "epoch": 5.89,
      "learning_rate": 3.0393667463955914e-06,
      "loss": 0.0164,
      "step": 6206,
      "task_loss": 0.004152687266469002
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981747197657181,
      "compression_loss": 0.0,
      "distillation_loss": 0.043606314808130264,
      "epoch": 5.89,
      "learning_rate": 3.034276816301693e-06,
      "loss": 0.0501,
      "step": 6207,
      "task_loss": 0.10881586372852325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981771874718745,
      "compression_loss": 0.0,
      "distillation_loss": 0.03161245957016945,
      "epoch": 5.9,
      "learning_rate": 3.0291908763872e-06,
      "loss": 0.0308,
      "step": 6208,
      "task_loss": 0.023898892104625702
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981796529528676,
      "compression_loss": 0.0,
      "distillation_loss": 0.1580495983362198,
      "epoch": 5.9,
      "learning_rate": 3.0241089275759886e-06,
      "loss": 0.1547,
      "step": 6209,
      "task_loss": 0.12449827790260315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981821162097013,
      "compression_loss": 0.0,
      "distillation_loss": 0.017794454470276833,
      "epoch": 5.9,
      "learning_rate": 3.019030970791237e-06,
      "loss": 0.0165,
      "step": 6210,
      "task_loss": 0.004612291231751442
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981845772433791,
      "compression_loss": 0.0,
      "distillation_loss": 0.06614623963832855,
      "epoch": 5.9,
      "learning_rate": 3.013957006955362e-06,
      "loss": 0.0748,
      "step": 6211,
      "task_loss": 0.1525183916091919
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981870360549047,
      "compression_loss": 0.0,
      "distillation_loss": 0.10029789060354233,
      "epoch": 5.9,
      "learning_rate": 3.008887036990085e-06,
      "loss": 0.0955,
      "step": 6212,
      "task_loss": 0.05255453288555145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981894926452819,
      "compression_loss": 0.0,
      "distillation_loss": 0.08173903077840805,
      "epoch": 5.9,
      "learning_rate": 3.003821061816392e-06,
      "loss": 0.0937,
      "step": 6213,
      "task_loss": 0.20181068778038025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981919470155143,
      "compression_loss": 0.0,
      "distillation_loss": 0.02389482408761978,
      "epoch": 5.9,
      "learning_rate": 2.998759082354538e-06,
      "loss": 0.0471,
      "step": 6214,
      "task_loss": 0.25638294219970703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981943991666055,
      "compression_loss": 0.0,
      "distillation_loss": 0.060066595673561096,
      "epoch": 5.9,
      "learning_rate": 2.9937010995240543e-06,
      "loss": 0.0666,
      "step": 6215,
      "task_loss": 0.12505532801151276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981968490995595,
      "compression_loss": 0.0,
      "distillation_loss": 0.03829333186149597,
      "epoch": 5.9,
      "learning_rate": 2.988647114243748e-06,
      "loss": 0.0417,
      "step": 6216,
      "task_loss": 0.0723431333899498
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7981992968153795,
      "compression_loss": 0.0,
      "distillation_loss": 0.06221352517604828,
      "epoch": 5.9,
      "learning_rate": 2.983597127431703e-06,
      "loss": 0.0646,
      "step": 6217,
      "task_loss": 0.08588936179876328
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982017423150696,
      "compression_loss": 0.0,
      "distillation_loss": 0.0338950976729393,
      "epoch": 5.91,
      "learning_rate": 2.9785511400052727e-06,
      "loss": 0.0451,
      "step": 6218,
      "task_loss": 0.14563745260238647
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982041855996332,
      "compression_loss": 0.0,
      "distillation_loss": 0.10232432931661606,
      "epoch": 5.91,
      "learning_rate": 2.973509152881079e-06,
      "loss": 0.1044,
      "step": 6219,
      "task_loss": 0.12259991466999054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982066266700742,
      "compression_loss": 0.0,
      "distillation_loss": 0.043488338589668274,
      "epoch": 5.91,
      "learning_rate": 2.9684711669750313e-06,
      "loss": 0.0447,
      "step": 6220,
      "task_loss": 0.05578209087252617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982090655273961,
      "compression_loss": 0.0,
      "distillation_loss": 0.02594229206442833,
      "epoch": 5.91,
      "learning_rate": 2.963437183202303e-06,
      "loss": 0.0315,
      "step": 6221,
      "task_loss": 0.08114132285118103
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982115021726028,
      "compression_loss": 0.0,
      "distillation_loss": 0.03786736726760864,
      "epoch": 5.91,
      "learning_rate": 2.9584072024773395e-06,
      "loss": 0.0364,
      "step": 6222,
      "task_loss": 0.022759338840842247
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982139366066978,
      "compression_loss": 0.0,
      "distillation_loss": 0.01899906061589718,
      "epoch": 5.91,
      "learning_rate": 2.953381225713858e-06,
      "loss": 0.0281,
      "step": 6223,
      "task_loss": 0.110114686191082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982163688306848,
      "compression_loss": 0.0,
      "distillation_loss": 0.07625134289264679,
      "epoch": 5.91,
      "learning_rate": 2.948359253824859e-06,
      "loss": 0.0865,
      "step": 6224,
      "task_loss": 0.1790730506181717
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982187988455675,
      "compression_loss": 0.0,
      "distillation_loss": 0.04318426176905632,
      "epoch": 5.91,
      "learning_rate": 2.9433412877226036e-06,
      "loss": 0.0521,
      "step": 6225,
      "task_loss": 0.1328027844429016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982212266523496,
      "compression_loss": 0.0,
      "distillation_loss": 0.11644196510314941,
      "epoch": 5.91,
      "learning_rate": 2.938327328318624e-06,
      "loss": 0.1194,
      "step": 6226,
      "task_loss": 0.14612026512622833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982236522520348,
      "compression_loss": 0.0,
      "distillation_loss": 0.04620961844921112,
      "epoch": 5.91,
      "learning_rate": 2.9333173765237394e-06,
      "loss": 0.0492,
      "step": 6227,
      "task_loss": 0.07610806822776794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982260756456268,
      "compression_loss": 0.0,
      "distillation_loss": 0.02641240693628788,
      "epoch": 5.91,
      "learning_rate": 2.928311433248024e-06,
      "loss": 0.025,
      "step": 6228,
      "task_loss": 0.012464674189686775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982284968341291,
      "compression_loss": 0.0,
      "distillation_loss": 0.04388430714607239,
      "epoch": 5.92,
      "learning_rate": 2.923309499400839e-06,
      "loss": 0.0429,
      "step": 6229,
      "task_loss": 0.03393597900867462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982309158185457,
      "compression_loss": 0.0,
      "distillation_loss": 0.0793541669845581,
      "epoch": 5.92,
      "learning_rate": 2.918311575890803e-06,
      "loss": 0.0829,
      "step": 6230,
      "task_loss": 0.11435024440288544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982333325998799,
      "compression_loss": 0.0,
      "distillation_loss": 0.0714527815580368,
      "epoch": 5.92,
      "learning_rate": 2.9133176636258196e-06,
      "loss": 0.0659,
      "step": 6231,
      "task_loss": 0.015902981162071228
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982357471791358,
      "compression_loss": 0.0,
      "distillation_loss": 0.07313866913318634,
      "epoch": 5.92,
      "learning_rate": 2.9083277635130523e-06,
      "loss": 0.0717,
      "step": 6232,
      "task_loss": 0.05869182199239731
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982381595573168,
      "compression_loss": 0.0,
      "distillation_loss": 0.010230525396764278,
      "epoch": 5.92,
      "learning_rate": 2.903341876458937e-06,
      "loss": 0.0096,
      "step": 6233,
      "task_loss": 0.004364604130387306
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982405697354266,
      "compression_loss": 0.0,
      "distillation_loss": 0.040943779051303864,
      "epoch": 5.92,
      "learning_rate": 2.8983600033691936e-06,
      "loss": 0.046,
      "step": 6234,
      "task_loss": 0.09106667339801788
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798242977714469,
      "compression_loss": 0.0,
      "distillation_loss": 0.08053984493017197,
      "epoch": 5.92,
      "learning_rate": 2.8933821451487987e-06,
      "loss": 0.0847,
      "step": 6235,
      "task_loss": 0.12223599851131439
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982453834954475,
      "compression_loss": 0.0,
      "distillation_loss": 0.018065018579363823,
      "epoch": 5.92,
      "learning_rate": 2.8884083027019976e-06,
      "loss": 0.0273,
      "step": 6236,
      "task_loss": 0.11061768233776093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982477870793661,
      "compression_loss": 0.0,
      "distillation_loss": 0.049103301018476486,
      "epoch": 5.92,
      "learning_rate": 2.8834384769323207e-06,
      "loss": 0.0446,
      "step": 6237,
      "task_loss": 0.0044719260185956955
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982501884672281,
      "compression_loss": 0.0,
      "distillation_loss": 0.0168076790869236,
      "epoch": 5.92,
      "learning_rate": 2.878472668742563e-06,
      "loss": 0.0157,
      "step": 6238,
      "task_loss": 0.005568409338593483
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982525876600374,
      "compression_loss": 0.0,
      "distillation_loss": 0.14126557111740112,
      "epoch": 5.92,
      "learning_rate": 2.8735108790347824e-06,
      "loss": 0.1368,
      "step": 6239,
      "task_loss": 0.0968267098069191
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982549846587976,
      "compression_loss": 0.0,
      "distillation_loss": 0.028058458119630814,
      "epoch": 5.93,
      "learning_rate": 2.8685531087103092e-06,
      "loss": 0.0302,
      "step": 6240,
      "task_loss": 0.049860235303640366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982573794645126,
      "compression_loss": 0.0,
      "distillation_loss": 0.028030022978782654,
      "epoch": 5.93,
      "learning_rate": 2.8635993586697553e-06,
      "loss": 0.0262,
      "step": 6241,
      "task_loss": 0.010059693828225136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982597720781858,
      "compression_loss": 0.0,
      "distillation_loss": 0.10418315976858139,
      "epoch": 5.93,
      "learning_rate": 2.8586496298129873e-06,
      "loss": 0.1219,
      "step": 6242,
      "task_loss": 0.28089088201522827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982621625008209,
      "compression_loss": 0.0,
      "distillation_loss": 0.01780703291296959,
      "epoch": 5.93,
      "learning_rate": 2.8537039230391416e-06,
      "loss": 0.0163,
      "step": 6243,
      "task_loss": 0.003232475370168686
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982645507334218,
      "compression_loss": 0.0,
      "distillation_loss": 0.07454448938369751,
      "epoch": 5.93,
      "learning_rate": 2.8487622392466438e-06,
      "loss": 0.0806,
      "step": 6244,
      "task_loss": 0.13465920090675354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798266936776992,
      "compression_loss": 0.0,
      "distillation_loss": 0.025488190352916718,
      "epoch": 5.93,
      "learning_rate": 2.8438245793331598e-06,
      "loss": 0.0241,
      "step": 6245,
      "task_loss": 0.011609837412834167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982693206325353,
      "compression_loss": 0.0,
      "distillation_loss": 0.11189248412847519,
      "epoch": 5.93,
      "learning_rate": 2.8388909441956517e-06,
      "loss": 0.1093,
      "step": 6246,
      "task_loss": 0.08592940866947174
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982717023010553,
      "compression_loss": 0.0,
      "distillation_loss": 0.01671120896935463,
      "epoch": 5.93,
      "learning_rate": 2.8339613347303283e-06,
      "loss": 0.0154,
      "step": 6247,
      "task_loss": 0.003899950534105301
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982740817835556,
      "compression_loss": 0.0,
      "distillation_loss": 0.03453841432929039,
      "epoch": 5.93,
      "learning_rate": 2.8290357518326843e-06,
      "loss": 0.0494,
      "step": 6248,
      "task_loss": 0.18323814868927002
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982764590810402,
      "compression_loss": 0.0,
      "distillation_loss": 0.04547137767076492,
      "epoch": 5.93,
      "learning_rate": 2.824114196397476e-06,
      "loss": 0.0428,
      "step": 6249,
      "task_loss": 0.01905733346939087
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982788341945124,
      "compression_loss": 0.0,
      "distillation_loss": 0.02116478979587555,
      "epoch": 5.94,
      "learning_rate": 2.819196669318716e-06,
      "loss": 0.0196,
      "step": 6250,
      "task_loss": 0.0053593870252370834
    },
    {
      "epoch": 5.94,
      "eval_accuracy": 0.8841743119266054,
      "eval_loss": 0.43586060404777527,
      "eval_runtime": 17.6466,
      "eval_samples_per_second": 49.415,
      "eval_steps_per_second": 6.177,
      "step": 6250
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982812071249761,
      "compression_loss": 0.0,
      "distillation_loss": 0.013482020236551762,
      "epoch": 5.94,
      "learning_rate": 2.814283171489712e-06,
      "loss": 0.0127,
      "step": 6251,
      "task_loss": 0.0053385235369205475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982835778734348,
      "compression_loss": 0.0,
      "distillation_loss": 0.08214183896780014,
      "epoch": 5.94,
      "learning_rate": 2.809373703803017e-06,
      "loss": 0.0771,
      "step": 6252,
      "task_loss": 0.0316874198615551
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982859464408925,
      "compression_loss": 0.0,
      "distillation_loss": 0.03672255948185921,
      "epoch": 5.94,
      "learning_rate": 2.8044682671504536e-06,
      "loss": 0.0444,
      "step": 6253,
      "task_loss": 0.11331107467412949
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982883128283526,
      "compression_loss": 0.0,
      "distillation_loss": 0.08191946148872375,
      "epoch": 5.94,
      "learning_rate": 2.7995668624231253e-06,
      "loss": 0.0842,
      "step": 6254,
      "task_loss": 0.10507936030626297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982906770368189,
      "compression_loss": 0.0,
      "distillation_loss": 0.04623375087976456,
      "epoch": 5.94,
      "learning_rate": 2.7946694905114006e-06,
      "loss": 0.0583,
      "step": 6255,
      "task_loss": 0.16685155034065247
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798293039067295,
      "compression_loss": 0.0,
      "distillation_loss": 0.012130487710237503,
      "epoch": 5.94,
      "learning_rate": 2.789776152304904e-06,
      "loss": 0.0189,
      "step": 6256,
      "task_loss": 0.0798635482788086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982953989207847,
      "compression_loss": 0.0,
      "distillation_loss": 0.02477359026670456,
      "epoch": 5.94,
      "learning_rate": 2.7848868486925307e-06,
      "loss": 0.0283,
      "step": 6257,
      "task_loss": 0.0600624680519104
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7982977565982916,
      "compression_loss": 0.0,
      "distillation_loss": 0.022501779720187187,
      "epoch": 5.94,
      "learning_rate": 2.7800015805624526e-06,
      "loss": 0.0403,
      "step": 6258,
      "task_loss": 0.20015643537044525
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983001121008194,
      "compression_loss": 0.0,
      "distillation_loss": 0.011296769604086876,
      "epoch": 5.94,
      "learning_rate": 2.7751203488021042e-06,
      "loss": 0.019,
      "step": 6259,
      "task_loss": 0.08792918175458908
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983024654293718,
      "compression_loss": 0.0,
      "distillation_loss": 0.030689077451825142,
      "epoch": 5.94,
      "learning_rate": 2.7702431542981715e-06,
      "loss": 0.038,
      "step": 6260,
      "task_loss": 0.10409321635961533
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983048165849524,
      "compression_loss": 0.0,
      "distillation_loss": 0.07838670909404755,
      "epoch": 5.95,
      "learning_rate": 2.765369997936637e-06,
      "loss": 0.0724,
      "step": 6261,
      "task_loss": 0.018565570935606956
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983071655685651,
      "compression_loss": 0.0,
      "distillation_loss": 0.02527971751987934,
      "epoch": 5.95,
      "learning_rate": 2.7605008806027206e-06,
      "loss": 0.0422,
      "step": 6262,
      "task_loss": 0.19478943943977356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983095123812134,
      "compression_loss": 0.0,
      "distillation_loss": 0.013227314688265324,
      "epoch": 5.95,
      "learning_rate": 2.7556358031809308e-06,
      "loss": 0.0121,
      "step": 6263,
      "task_loss": 0.002236088737845421
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798311857023901,
      "compression_loss": 0.0,
      "distillation_loss": 0.013676393777132034,
      "epoch": 5.95,
      "learning_rate": 2.7507747665550227e-06,
      "loss": 0.014,
      "step": 6264,
      "task_loss": 0.01731892116367817
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983141994976315,
      "compression_loss": 0.0,
      "distillation_loss": 0.05857694894075394,
      "epoch": 5.95,
      "learning_rate": 2.7459177716080363e-06,
      "loss": 0.0554,
      "step": 6265,
      "task_loss": 0.02669447846710682
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983165398034088,
      "compression_loss": 0.0,
      "distillation_loss": 0.041833531111478806,
      "epoch": 5.95,
      "learning_rate": 2.741064819222264e-06,
      "loss": 0.0408,
      "step": 6266,
      "task_loss": 0.03170401230454445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983188779422363,
      "compression_loss": 0.0,
      "distillation_loss": 0.024519365280866623,
      "epoch": 5.95,
      "learning_rate": 2.7362159102792605e-06,
      "loss": 0.0226,
      "step": 6267,
      "task_loss": 0.005678284913301468
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798321213915118,
      "compression_loss": 0.0,
      "distillation_loss": 0.017905663698911667,
      "epoch": 5.95,
      "learning_rate": 2.7313710456598667e-06,
      "loss": 0.0166,
      "step": 6268,
      "task_loss": 0.004889186471700668
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983235477230574,
      "compression_loss": 0.0,
      "distillation_loss": 0.09741759300231934,
      "epoch": 5.95,
      "learning_rate": 2.72653022624417e-06,
      "loss": 0.0961,
      "step": 6269,
      "task_loss": 0.08380945026874542
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983258793670581,
      "compression_loss": 0.0,
      "distillation_loss": 0.028389321640133858,
      "epoch": 5.95,
      "learning_rate": 2.7216934529115223e-06,
      "loss": 0.0266,
      "step": 6270,
      "task_loss": 0.010113891214132309
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798328208848124,
      "compression_loss": 0.0,
      "distillation_loss": 0.012037597596645355,
      "epoch": 5.96,
      "learning_rate": 2.7168607265405517e-06,
      "loss": 0.0111,
      "step": 6271,
      "task_loss": 0.002512754872441292
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983305361672587,
      "compression_loss": 0.0,
      "distillation_loss": 0.016703153029084206,
      "epoch": 5.96,
      "learning_rate": 2.712032048009153e-06,
      "loss": 0.0232,
      "step": 6272,
      "task_loss": 0.08192671090364456
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983328613254658,
      "compression_loss": 0.0,
      "distillation_loss": 0.06296996772289276,
      "epoch": 5.96,
      "learning_rate": 2.707207418194474e-06,
      "loss": 0.0594,
      "step": 6273,
      "task_loss": 0.026859570294618607
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798335184323749,
      "compression_loss": 0.0,
      "distillation_loss": 0.0931231677532196,
      "epoch": 5.96,
      "learning_rate": 2.702386837972923e-06,
      "loss": 0.095,
      "step": 6274,
      "task_loss": 0.11223654448986053
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798337505163112,
      "compression_loss": 0.0,
      "distillation_loss": 0.08410904556512833,
      "epoch": 5.96,
      "learning_rate": 2.697570308220196e-06,
      "loss": 0.0936,
      "step": 6275,
      "task_loss": 0.17943312227725983
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983398238445586,
      "compression_loss": 0.0,
      "distillation_loss": 0.021530931815505028,
      "epoch": 5.96,
      "learning_rate": 2.6927578298112324e-06,
      "loss": 0.022,
      "step": 6276,
      "task_loss": 0.025872595608234406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983421403690923,
      "compression_loss": 0.0,
      "distillation_loss": 0.1349969059228897,
      "epoch": 5.96,
      "learning_rate": 2.687949403620235e-06,
      "loss": 0.1343,
      "step": 6277,
      "task_loss": 0.1277485191822052
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983444547377169,
      "compression_loss": 0.0,
      "distillation_loss": 0.047185495495796204,
      "epoch": 5.96,
      "learning_rate": 2.6831450305206896e-06,
      "loss": 0.044,
      "step": 6278,
      "task_loss": 0.015110282227396965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798346766951436,
      "compression_loss": 0.0,
      "distillation_loss": 0.028054438531398773,
      "epoch": 5.96,
      "learning_rate": 2.6783447113853305e-06,
      "loss": 0.026,
      "step": 6279,
      "task_loss": 0.00766693614423275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983490770112535,
      "compression_loss": 0.0,
      "distillation_loss": 0.01661674678325653,
      "epoch": 5.96,
      "learning_rate": 2.673548447086152e-06,
      "loss": 0.0274,
      "step": 6280,
      "task_loss": 0.12479189783334732
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983513849181727,
      "compression_loss": 0.0,
      "distillation_loss": 0.012853579595685005,
      "epoch": 5.96,
      "learning_rate": 2.6687562384944216e-06,
      "loss": 0.0221,
      "step": 6281,
      "task_loss": 0.10521620512008667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983536906731976,
      "compression_loss": 0.0,
      "distillation_loss": 0.032144561409950256,
      "epoch": 5.97,
      "learning_rate": 2.6639680864806754e-06,
      "loss": 0.0308,
      "step": 6282,
      "task_loss": 0.01860850676894188
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983559942773317,
      "compression_loss": 0.0,
      "distillation_loss": 0.023536494001746178,
      "epoch": 5.97,
      "learning_rate": 2.659183991914696e-06,
      "loss": 0.027,
      "step": 6283,
      "task_loss": 0.05826903134584427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983582957315788,
      "compression_loss": 0.0,
      "distillation_loss": 0.04953496530652046,
      "epoch": 5.97,
      "learning_rate": 2.6544039556655354e-06,
      "loss": 0.054,
      "step": 6284,
      "task_loss": 0.09401563555002213
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983605950369426,
      "compression_loss": 0.0,
      "distillation_loss": 0.011628339067101479,
      "epoch": 5.97,
      "learning_rate": 2.6496279786015184e-06,
      "loss": 0.0187,
      "step": 6285,
      "task_loss": 0.08250655233860016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983628921944266,
      "compression_loss": 0.0,
      "distillation_loss": 0.03543800860643387,
      "epoch": 5.97,
      "learning_rate": 2.6448560615902202e-06,
      "loss": 0.0324,
      "step": 6286,
      "task_loss": 0.0052569154649972916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983651872050347,
      "compression_loss": 0.0,
      "distillation_loss": 0.015366720035672188,
      "epoch": 5.97,
      "learning_rate": 2.640088205498481e-06,
      "loss": 0.0165,
      "step": 6287,
      "task_loss": 0.0262142401188612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983674800697704,
      "compression_loss": 0.0,
      "distillation_loss": 0.04069266468286514,
      "epoch": 5.97,
      "learning_rate": 2.635324411192397e-06,
      "loss": 0.042,
      "step": 6288,
      "task_loss": 0.054010894149541855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983697707896376,
      "compression_loss": 0.0,
      "distillation_loss": 0.08131635189056396,
      "epoch": 5.97,
      "learning_rate": 2.6305646795373507e-06,
      "loss": 0.0923,
      "step": 6289,
      "task_loss": 0.19130820035934448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983720593656397,
      "compression_loss": 0.0,
      "distillation_loss": 0.022895563393831253,
      "epoch": 5.97,
      "learning_rate": 2.6258090113979646e-06,
      "loss": 0.0363,
      "step": 6290,
      "task_loss": 0.15735076367855072
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983743457987806,
      "compression_loss": 0.0,
      "distillation_loss": 0.018831467255949974,
      "epoch": 5.97,
      "learning_rate": 2.6210574076381224e-06,
      "loss": 0.0432,
      "step": 6291,
      "task_loss": 0.26204773783683777
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798376630090064,
      "compression_loss": 0.0,
      "distillation_loss": 0.0476510114967823,
      "epoch": 5.98,
      "learning_rate": 2.616309869120984e-06,
      "loss": 0.0463,
      "step": 6292,
      "task_loss": 0.033927060663700104
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983789122404933,
      "compression_loss": 0.0,
      "distillation_loss": 0.04823393374681473,
      "epoch": 5.98,
      "learning_rate": 2.6115663967089587e-06,
      "loss": 0.0457,
      "step": 6293,
      "task_loss": 0.022519726306200027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983811922510725,
      "compression_loss": 0.0,
      "distillation_loss": 0.02489767223596573,
      "epoch": 5.98,
      "learning_rate": 2.6068269912637224e-06,
      "loss": 0.0393,
      "step": 6294,
      "task_loss": 0.16874471306800842
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983834701228052,
      "compression_loss": 0.0,
      "distillation_loss": 0.01917928084731102,
      "epoch": 5.98,
      "learning_rate": 2.602091653646205e-06,
      "loss": 0.0256,
      "step": 6295,
      "task_loss": 0.08345439285039902
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798385745856695,
      "compression_loss": 0.0,
      "distillation_loss": 0.011729862540960312,
      "epoch": 5.98,
      "learning_rate": 2.597360384716613e-06,
      "loss": 0.0111,
      "step": 6296,
      "task_loss": 0.00561935268342495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983880194537456,
      "compression_loss": 0.0,
      "distillation_loss": 0.08482478559017181,
      "epoch": 5.98,
      "learning_rate": 2.592633185334395e-06,
      "loss": 0.0894,
      "step": 6297,
      "task_loss": 0.1304791271686554
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983902909149607,
      "compression_loss": 0.0,
      "distillation_loss": 0.0635722205042839,
      "epoch": 5.98,
      "learning_rate": 2.5879100563582724e-06,
      "loss": 0.0594,
      "step": 6298,
      "task_loss": 0.022111881524324417
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798392560241344,
      "compression_loss": 0.0,
      "distillation_loss": 0.023897292092442513,
      "epoch": 5.98,
      "learning_rate": 2.5831909986462318e-06,
      "loss": 0.0219,
      "step": 6299,
      "task_loss": 0.004307908937335014
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983948274338992,
      "compression_loss": 0.0,
      "distillation_loss": 0.024204891175031662,
      "epoch": 5.98,
      "learning_rate": 2.5784760130555057e-06,
      "loss": 0.0237,
      "step": 6300,
      "task_loss": 0.018668916076421738
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983970924936299,
      "compression_loss": 0.0,
      "distillation_loss": 0.018057633191347122,
      "epoch": 5.98,
      "learning_rate": 2.573765100442596e-06,
      "loss": 0.0167,
      "step": 6301,
      "task_loss": 0.004313705489039421
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7983993554215398,
      "compression_loss": 0.0,
      "distillation_loss": 0.05510625243186951,
      "epoch": 5.98,
      "learning_rate": 2.569058261663257e-06,
      "loss": 0.0541,
      "step": 6302,
      "task_loss": 0.04473312571644783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984016162186327,
      "compression_loss": 0.0,
      "distillation_loss": 0.016285603865981102,
      "epoch": 5.99,
      "learning_rate": 2.5643554975725182e-06,
      "loss": 0.0216,
      "step": 6303,
      "task_loss": 0.0692404955625534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984038748859122,
      "compression_loss": 0.0,
      "distillation_loss": 0.021900523453950882,
      "epoch": 5.99,
      "learning_rate": 2.5596568090246548e-06,
      "loss": 0.0205,
      "step": 6304,
      "task_loss": 0.007533866912126541
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798406131424382,
      "compression_loss": 0.0,
      "distillation_loss": 0.0129203200340271,
      "epoch": 5.99,
      "learning_rate": 2.5549621968732005e-06,
      "loss": 0.0236,
      "step": 6305,
      "task_loss": 0.12000947445631027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984083858350457,
      "compression_loss": 0.0,
      "distillation_loss": 0.02725234068930149,
      "epoch": 5.99,
      "learning_rate": 2.5502716619709598e-06,
      "loss": 0.0256,
      "step": 6306,
      "task_loss": 0.01108565554022789
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984106381189071,
      "compression_loss": 0.0,
      "distillation_loss": 0.03494996577501297,
      "epoch": 5.99,
      "learning_rate": 2.545585205169995e-06,
      "loss": 0.0337,
      "step": 6307,
      "task_loss": 0.022902732715010643
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984128882769698,
      "compression_loss": 0.0,
      "distillation_loss": 0.026792103424668312,
      "epoch": 5.99,
      "learning_rate": 2.54090282732162e-06,
      "loss": 0.0306,
      "step": 6308,
      "task_loss": 0.06438016891479492
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984151363102374,
      "compression_loss": 0.0,
      "distillation_loss": 0.015818610787391663,
      "epoch": 5.99,
      "learning_rate": 2.536224529276404e-06,
      "loss": 0.0239,
      "step": 6309,
      "task_loss": 0.09637781977653503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984173822197138,
      "compression_loss": 0.0,
      "distillation_loss": 0.015181312337517738,
      "epoch": 5.99,
      "learning_rate": 2.5315503118841956e-06,
      "loss": 0.0272,
      "step": 6310,
      "task_loss": 0.1348828375339508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984196260064026,
      "compression_loss": 0.0,
      "distillation_loss": 0.021186668425798416,
      "epoch": 5.99,
      "learning_rate": 2.5268801759940813e-06,
      "loss": 0.02,
      "step": 6311,
      "task_loss": 0.009767893701791763
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984218676713074,
      "compression_loss": 0.0,
      "distillation_loss": 0.058261334896087646,
      "epoch": 5.99,
      "learning_rate": 2.522214122454411e-06,
      "loss": 0.0564,
      "step": 6312,
      "task_loss": 0.04001392796635628
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798424107215432,
      "compression_loss": 0.0,
      "distillation_loss": 0.02801283821463585,
      "epoch": 6.0,
      "learning_rate": 2.5175521521128035e-06,
      "loss": 0.0276,
      "step": 6313,
      "task_loss": 0.023807687684893608
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79842634463978,
      "compression_loss": 0.0,
      "distillation_loss": 0.166382297873497,
      "epoch": 6.0,
      "learning_rate": 2.512894265816121e-06,
      "loss": 0.165,
      "step": 6314,
      "task_loss": 0.15275904536247253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984285799453551,
      "compression_loss": 0.0,
      "distillation_loss": 0.06544499844312668,
      "epoch": 6.0,
      "learning_rate": 2.5082404644104995e-06,
      "loss": 0.0815,
      "step": 6315,
      "task_loss": 0.22553624212741852
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984308131331609,
      "compression_loss": 0.0,
      "distillation_loss": 0.05184311419725418,
      "epoch": 6.0,
      "learning_rate": 2.503590748741311e-06,
      "loss": 0.0555,
      "step": 6316,
      "task_loss": 0.0887538492679596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984330442042012,
      "compression_loss": 0.0,
      "distillation_loss": 0.023621458560228348,
      "epoch": 6.0,
      "learning_rate": 2.498945119653215e-06,
      "loss": 0.0217,
      "step": 6317,
      "task_loss": 0.00461687333881855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.620820711904918,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7947265660077046,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984352731594797,
      "compression_loss": 0.0,
      "distillation_loss": 0.03740524873137474,
      "epoch": 6.0,
      "learning_rate": 2.4943035779901008e-06,
      "loss": 0.0582,
      "step": 6318,
      "task_loss": 0.24517786502838135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984375,
      "compression_loss": 0.0,
      "distillation_loss": 0.23347151279449463,
      "epoch": 6.0,
      "learning_rate": 2.4896661245951275e-06,
      "loss": 0.2215,
      "step": 6319,
      "task_loss": 0.11372692137956619
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984397247267658,
      "compression_loss": 0.0,
      "distillation_loss": 0.3618473410606384,
      "epoch": 6.0,
      "learning_rate": 2.485032760310718e-06,
      "loss": 0.3399,
      "step": 6320,
      "task_loss": 0.14205172657966614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984419473407809,
      "compression_loss": 0.0,
      "distillation_loss": 0.2971758544445038,
      "epoch": 6.0,
      "learning_rate": 2.4804034859785414e-06,
      "loss": 0.2839,
      "step": 6321,
      "task_loss": 0.16441358625888824
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984441678430487,
      "compression_loss": 0.0,
      "distillation_loss": 0.21649004518985748,
      "epoch": 6.0,
      "learning_rate": 2.475778302439524e-06,
      "loss": 0.2025,
      "step": 6322,
      "task_loss": 0.07684586942195892
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984463862345732,
      "compression_loss": 0.0,
      "distillation_loss": 0.23935849964618683,
      "epoch": 6.0,
      "learning_rate": 2.4711572105338533e-06,
      "loss": 0.2226,
      "step": 6323,
      "task_loss": 0.07181292772293091
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984486025163579,
      "compression_loss": 0.0,
      "distillation_loss": 0.24455556273460388,
      "epoch": 6.01,
      "learning_rate": 2.4665402111009814e-06,
      "loss": 0.2304,
      "step": 6324,
      "task_loss": 0.10250243544578552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984508166894065,
      "compression_loss": 0.0,
      "distillation_loss": 0.3571939468383789,
      "epoch": 6.01,
      "learning_rate": 2.4619273049796e-06,
      "loss": 0.3339,
      "step": 6325,
      "task_loss": 0.12418486177921295
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984530287547227,
      "compression_loss": 0.0,
      "distillation_loss": 0.22672238945960999,
      "epoch": 6.01,
      "learning_rate": 2.4573184930076655e-06,
      "loss": 0.2185,
      "step": 6326,
      "task_loss": 0.14480708539485931
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984552387133103,
      "compression_loss": 0.0,
      "distillation_loss": 0.2425723820924759,
      "epoch": 6.01,
      "learning_rate": 2.4527137760224e-06,
      "loss": 0.2238,
      "step": 6327,
      "task_loss": 0.05520417168736458
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984574465661727,
      "compression_loss": 0.0,
      "distillation_loss": 0.2952827215194702,
      "epoch": 6.01,
      "learning_rate": 2.4481131548602627e-06,
      "loss": 0.2822,
      "step": 6328,
      "task_loss": 0.16485857963562012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984596523143138,
      "compression_loss": 0.0,
      "distillation_loss": 0.24353119730949402,
      "epoch": 6.01,
      "learning_rate": 2.443516630356979e-06,
      "loss": 0.2293,
      "step": 6329,
      "task_loss": 0.10098670423030853
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984618559587373,
      "compression_loss": 0.0,
      "distillation_loss": 0.2624762952327728,
      "epoch": 6.01,
      "learning_rate": 2.4389242033475366e-06,
      "loss": 0.2415,
      "step": 6330,
      "task_loss": 0.05250587314367294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984640575004468,
      "compression_loss": 0.0,
      "distillation_loss": 0.20973241329193115,
      "epoch": 6.01,
      "learning_rate": 2.4343358746661686e-06,
      "loss": 0.1955,
      "step": 6331,
      "task_loss": 0.06739503890275955
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984662569404459,
      "compression_loss": 0.0,
      "distillation_loss": 0.26178908348083496,
      "epoch": 6.01,
      "learning_rate": 2.4297516451463608e-06,
      "loss": 0.2473,
      "step": 6332,
      "task_loss": 0.11708267778158188
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984684542797384,
      "compression_loss": 0.0,
      "distillation_loss": 0.18669354915618896,
      "epoch": 6.01,
      "learning_rate": 2.4251715156208665e-06,
      "loss": 0.1795,
      "step": 6333,
      "task_loss": 0.11463765054941177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984706495193281,
      "compression_loss": 0.0,
      "distillation_loss": 0.305093377828598,
      "epoch": 6.02,
      "learning_rate": 2.4205954869216922e-06,
      "loss": 0.289,
      "step": 6334,
      "task_loss": 0.14380815625190735
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984728426602185,
      "compression_loss": 0.0,
      "distillation_loss": 0.15407568216323853,
      "epoch": 6.02,
      "learning_rate": 2.416023559880093e-06,
      "loss": 0.1493,
      "step": 6335,
      "task_loss": 0.1060083881020546
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984750337034132,
      "compression_loss": 0.0,
      "distillation_loss": 0.20316243171691895,
      "epoch": 6.02,
      "learning_rate": 2.4114557353265733e-06,
      "loss": 0.1901,
      "step": 6336,
      "task_loss": 0.0729086771607399
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798477222649916,
      "compression_loss": 0.0,
      "distillation_loss": 0.26141518354415894,
      "epoch": 6.02,
      "learning_rate": 2.4068920140909114e-06,
      "loss": 0.2509,
      "step": 6337,
      "task_loss": 0.15607582032680511
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984794095007308,
      "compression_loss": 0.0,
      "distillation_loss": 0.22481253743171692,
      "epoch": 6.02,
      "learning_rate": 2.4023323970021273e-06,
      "loss": 0.2087,
      "step": 6338,
      "task_loss": 0.06392265856266022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984815942568609,
      "compression_loss": 0.0,
      "distillation_loss": 0.16369011998176575,
      "epoch": 6.02,
      "learning_rate": 2.397776884888489e-06,
      "loss": 0.1505,
      "step": 6339,
      "task_loss": 0.03216647729277611
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984837769193103,
      "compression_loss": 0.0,
      "distillation_loss": 0.22370657324790955,
      "epoch": 6.02,
      "learning_rate": 2.393225478577532e-06,
      "loss": 0.2189,
      "step": 6340,
      "task_loss": 0.17515668272972107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984859574890824,
      "compression_loss": 0.0,
      "distillation_loss": 0.22182436287403107,
      "epoch": 6.02,
      "learning_rate": 2.3886781788960477e-06,
      "loss": 0.2059,
      "step": 6341,
      "task_loss": 0.06226883456110954
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798488135967181,
      "compression_loss": 0.0,
      "distillation_loss": 0.19668897986412048,
      "epoch": 6.02,
      "learning_rate": 2.384134986670067e-06,
      "loss": 0.1828,
      "step": 6342,
      "task_loss": 0.0580269992351532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984903123546099,
      "compression_loss": 0.0,
      "distillation_loss": 0.16411647200584412,
      "epoch": 6.02,
      "learning_rate": 2.3795959027248847e-06,
      "loss": 0.1578,
      "step": 6343,
      "task_loss": 0.10057196021080017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984924866523726,
      "compression_loss": 0.0,
      "distillation_loss": 0.21145261824131012,
      "epoch": 6.02,
      "learning_rate": 2.3750609278850505e-06,
      "loss": 0.1984,
      "step": 6344,
      "task_loss": 0.08086170256137848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984946588614729,
      "compression_loss": 0.0,
      "distillation_loss": 0.12663835287094116,
      "epoch": 6.03,
      "learning_rate": 2.3705300629743606e-06,
      "loss": 0.1277,
      "step": 6345,
      "task_loss": 0.13693031668663025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984968289829144,
      "compression_loss": 0.0,
      "distillation_loss": 0.19469054043293,
      "epoch": 6.03,
      "learning_rate": 2.3660033088158647e-06,
      "loss": 0.179,
      "step": 6346,
      "task_loss": 0.037706997245550156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7984989970177009,
      "compression_loss": 0.0,
      "distillation_loss": 0.16001702845096588,
      "epoch": 6.03,
      "learning_rate": 2.36148066623188e-06,
      "loss": 0.1503,
      "step": 6347,
      "task_loss": 0.06241552159190178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798501162966836,
      "compression_loss": 0.0,
      "distillation_loss": 0.2323281466960907,
      "epoch": 6.03,
      "learning_rate": 2.356962136043961e-06,
      "loss": 0.2186,
      "step": 6348,
      "task_loss": 0.09503776580095291
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985033268313234,
      "compression_loss": 0.0,
      "distillation_loss": 0.18243101239204407,
      "epoch": 6.03,
      "learning_rate": 2.3524477190729144e-06,
      "loss": 0.1675,
      "step": 6349,
      "task_loss": 0.03352750092744827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985054886121667,
      "compression_loss": 0.0,
      "distillation_loss": 0.24202725291252136,
      "epoch": 6.03,
      "learning_rate": 2.3479374161388124e-06,
      "loss": 0.2333,
      "step": 6350,
      "task_loss": 0.1544978767633438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985076483103698,
      "compression_loss": 0.0,
      "distillation_loss": 0.2221883088350296,
      "epoch": 6.03,
      "learning_rate": 2.343431228060977e-06,
      "loss": 0.2053,
      "step": 6351,
      "task_loss": 0.05342654138803482
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985098059269361,
      "compression_loss": 0.0,
      "distillation_loss": 0.17091111838817596,
      "epoch": 6.03,
      "learning_rate": 2.3389291556579732e-06,
      "loss": 0.1579,
      "step": 6352,
      "task_loss": 0.0405244342982769
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985119614628695,
      "compression_loss": 0.0,
      "distillation_loss": 0.25911980867385864,
      "epoch": 6.03,
      "learning_rate": 2.334431199747622e-06,
      "loss": 0.2423,
      "step": 6353,
      "task_loss": 0.09104666113853455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985141149191736,
      "compression_loss": 0.0,
      "distillation_loss": 0.09606660157442093,
      "epoch": 6.03,
      "learning_rate": 2.3299373611470053e-06,
      "loss": 0.0913,
      "step": 6354,
      "task_loss": 0.04837304726243019
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798516266296852,
      "compression_loss": 0.0,
      "distillation_loss": 0.15584638714790344,
      "epoch": 6.04,
      "learning_rate": 2.3254476406724483e-06,
      "loss": 0.1464,
      "step": 6355,
      "task_loss": 0.06180054694414139
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985184155969085,
      "compression_loss": 0.0,
      "distillation_loss": 0.11610227823257446,
      "epoch": 6.04,
      "learning_rate": 2.3209620391395236e-06,
      "loss": 0.1073,
      "step": 6356,
      "task_loss": 0.027644779533147812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985205628203469,
      "compression_loss": 0.0,
      "distillation_loss": 0.16248267889022827,
      "epoch": 6.04,
      "learning_rate": 2.316480557363071e-06,
      "loss": 0.1549,
      "step": 6357,
      "task_loss": 0.08620963990688324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985227079681706,
      "compression_loss": 0.0,
      "distillation_loss": 0.14060789346694946,
      "epoch": 6.04,
      "learning_rate": 2.3120031961571697e-06,
      "loss": 0.1372,
      "step": 6358,
      "task_loss": 0.10676179826259613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985248510413834,
      "compression_loss": 0.0,
      "distillation_loss": 0.09738312661647797,
      "epoch": 6.04,
      "learning_rate": 2.3075299563351554e-06,
      "loss": 0.0958,
      "step": 6359,
      "task_loss": 0.08182113617658615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798526992040989,
      "compression_loss": 0.0,
      "distillation_loss": 0.11283515393733978,
      "epoch": 6.04,
      "learning_rate": 2.3030608387096087e-06,
      "loss": 0.1166,
      "step": 6360,
      "task_loss": 0.1502501219511032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798529130967991,
      "compression_loss": 0.0,
      "distillation_loss": 0.1467873603105545,
      "epoch": 6.04,
      "learning_rate": 2.298595844092377e-06,
      "loss": 0.1372,
      "step": 6361,
      "task_loss": 0.05081057548522949
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985312678233933,
      "compression_loss": 0.0,
      "distillation_loss": 0.14234556257724762,
      "epoch": 6.04,
      "learning_rate": 2.29413497329454e-06,
      "loss": 0.138,
      "step": 6362,
      "task_loss": 0.09874287247657776
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985334026081995,
      "compression_loss": 0.0,
      "distillation_loss": 0.14752641320228577,
      "epoch": 6.04,
      "learning_rate": 2.289678227126432e-06,
      "loss": 0.1442,
      "step": 6363,
      "task_loss": 0.11388550698757172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798535535323413,
      "compression_loss": 0.0,
      "distillation_loss": 0.11193514615297318,
      "epoch": 6.04,
      "learning_rate": 2.2852256063976537e-06,
      "loss": 0.1022,
      "step": 6364,
      "task_loss": 0.014793351292610168
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798537665970038,
      "compression_loss": 0.0,
      "distillation_loss": 0.08797906339168549,
      "epoch": 6.04,
      "learning_rate": 2.28077711191704e-06,
      "loss": 0.0801,
      "step": 6365,
      "task_loss": 0.009286869317293167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985397945490776,
      "compression_loss": 0.0,
      "distillation_loss": 0.09950749576091766,
      "epoch": 6.05,
      "learning_rate": 2.276332744492676e-06,
      "loss": 0.0956,
      "step": 6366,
      "task_loss": 0.06009618192911148
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985419210615359,
      "compression_loss": 0.0,
      "distillation_loss": 0.1407734453678131,
      "epoch": 6.05,
      "learning_rate": 2.271892504931905e-06,
      "loss": 0.1377,
      "step": 6367,
      "task_loss": 0.1104123517870903
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985440455084164,
      "compression_loss": 0.0,
      "distillation_loss": 0.12445957213640213,
      "epoch": 6.05,
      "learning_rate": 2.2674563940413283e-06,
      "loss": 0.1133,
      "step": 6368,
      "task_loss": 0.012716732919216156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798546167890723,
      "compression_loss": 0.0,
      "distillation_loss": 0.13971540331840515,
      "epoch": 6.05,
      "learning_rate": 2.263024412626777e-06,
      "loss": 0.1393,
      "step": 6369,
      "task_loss": 0.13585472106933594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798548288209459,
      "compression_loss": 0.0,
      "distillation_loss": 0.09350334107875824,
      "epoch": 6.05,
      "learning_rate": 2.2585965614933395e-06,
      "loss": 0.089,
      "step": 6370,
      "task_loss": 0.048799920827150345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985504064656284,
      "compression_loss": 0.0,
      "distillation_loss": 0.13269993662834167,
      "epoch": 6.05,
      "learning_rate": 2.254172841445365e-06,
      "loss": 0.1258,
      "step": 6371,
      "task_loss": 0.06413581967353821
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985525226602348,
      "compression_loss": 0.0,
      "distillation_loss": 0.11976936459541321,
      "epoch": 6.05,
      "learning_rate": 2.249753253286441e-06,
      "loss": 0.1226,
      "step": 6372,
      "task_loss": 0.14833396673202515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985546367942818,
      "compression_loss": 0.0,
      "distillation_loss": 0.11396154761314392,
      "epoch": 6.05,
      "learning_rate": 2.2453377978194024e-06,
      "loss": 0.1091,
      "step": 6373,
      "task_loss": 0.06537455320358276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985567488687731,
      "compression_loss": 0.0,
      "distillation_loss": 0.10704049468040466,
      "epoch": 6.05,
      "learning_rate": 2.2409264758463363e-06,
      "loss": 0.0984,
      "step": 6374,
      "task_loss": 0.02111482247710228
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985588588847126,
      "compression_loss": 0.0,
      "distillation_loss": 0.1097719818353653,
      "epoch": 6.05,
      "learning_rate": 2.2365192881685843e-06,
      "loss": 0.1024,
      "step": 6375,
      "task_loss": 0.03654416650533676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985609668431037,
      "compression_loss": 0.0,
      "distillation_loss": 0.10831979662179947,
      "epoch": 6.06,
      "learning_rate": 2.232116235586737e-06,
      "loss": 0.1068,
      "step": 6376,
      "task_loss": 0.0928025022149086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985630727449502,
      "compression_loss": 0.0,
      "distillation_loss": 0.0745440348982811,
      "epoch": 6.06,
      "learning_rate": 2.2277173189006266e-06,
      "loss": 0.0728,
      "step": 6377,
      "task_loss": 0.05667824670672417
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985651765912558,
      "compression_loss": 0.0,
      "distillation_loss": 0.09890006482601166,
      "epoch": 6.06,
      "learning_rate": 2.223322538909339e-06,
      "loss": 0.0972,
      "step": 6378,
      "task_loss": 0.0820925384759903
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985672783830241,
      "compression_loss": 0.0,
      "distillation_loss": 0.06193868815898895,
      "epoch": 6.06,
      "learning_rate": 2.2189318964112054e-06,
      "loss": 0.0637,
      "step": 6379,
      "task_loss": 0.07989262044429779
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985693781212589,
      "compression_loss": 0.0,
      "distillation_loss": 0.09831628203392029,
      "epoch": 6.06,
      "learning_rate": 2.2145453922038093e-06,
      "loss": 0.1028,
      "step": 6380,
      "task_loss": 0.14268594980239868
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985714758069639,
      "compression_loss": 0.0,
      "distillation_loss": 0.1020318865776062,
      "epoch": 6.06,
      "learning_rate": 2.210163027083975e-06,
      "loss": 0.0982,
      "step": 6381,
      "task_loss": 0.06365399807691574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985735714411425,
      "compression_loss": 0.0,
      "distillation_loss": 0.12122917175292969,
      "epoch": 6.06,
      "learning_rate": 2.205784801847785e-06,
      "loss": 0.1198,
      "step": 6382,
      "task_loss": 0.10688640177249908
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985756650247987,
      "compression_loss": 0.0,
      "distillation_loss": 0.11438720673322678,
      "epoch": 6.06,
      "learning_rate": 2.2014107172905666e-06,
      "loss": 0.1126,
      "step": 6383,
      "task_loss": 0.0968388095498085
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798577756558936,
      "compression_loss": 0.0,
      "distillation_loss": 0.15790635347366333,
      "epoch": 6.06,
      "learning_rate": 2.197040774206882e-06,
      "loss": 0.1469,
      "step": 6384,
      "task_loss": 0.0477466806769371
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985798460445582,
      "compression_loss": 0.0,
      "distillation_loss": 0.16549362242221832,
      "epoch": 6.06,
      "learning_rate": 2.192674973390568e-06,
      "loss": 0.1518,
      "step": 6385,
      "task_loss": 0.028843212872743607
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985819334826689,
      "compression_loss": 0.0,
      "distillation_loss": 0.08308826386928558,
      "epoch": 6.06,
      "learning_rate": 2.188313315634688e-06,
      "loss": 0.0802,
      "step": 6386,
      "task_loss": 0.05457150936126709
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985840188742719,
      "compression_loss": 0.0,
      "distillation_loss": 0.10413433611392975,
      "epoch": 6.07,
      "learning_rate": 2.1839558017315547e-06,
      "loss": 0.0995,
      "step": 6387,
      "task_loss": 0.057557813823223114
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985861022203707,
      "compression_loss": 0.0,
      "distillation_loss": 0.0583919882774353,
      "epoch": 6.07,
      "learning_rate": 2.1796024324727297e-06,
      "loss": 0.0768,
      "step": 6388,
      "task_loss": 0.24291850626468658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985881835219691,
      "compression_loss": 0.0,
      "distillation_loss": 0.07095431536436081,
      "epoch": 6.07,
      "learning_rate": 2.1752532086490333e-06,
      "loss": 0.0648,
      "step": 6389,
      "task_loss": 0.009517394006252289
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985902627800708,
      "compression_loss": 0.0,
      "distillation_loss": 0.13259541988372803,
      "epoch": 6.07,
      "learning_rate": 2.1709081310505143e-06,
      "loss": 0.1337,
      "step": 6390,
      "task_loss": 0.14401474595069885
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985923399956794,
      "compression_loss": 0.0,
      "distillation_loss": 0.06342358887195587,
      "epoch": 6.07,
      "learning_rate": 2.1665672004664765e-06,
      "loss": 0.0617,
      "step": 6391,
      "task_loss": 0.04634593799710274
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985944151697987,
      "compression_loss": 0.0,
      "distillation_loss": 0.14910653233528137,
      "epoch": 6.07,
      "learning_rate": 2.1622304176854736e-06,
      "loss": 0.1597,
      "step": 6392,
      "task_loss": 0.25551342964172363
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985964883034322,
      "compression_loss": 0.0,
      "distillation_loss": 0.054450199007987976,
      "epoch": 6.07,
      "learning_rate": 2.1578977834953053e-06,
      "loss": 0.05,
      "step": 6393,
      "task_loss": 0.010282203555107117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7985985593975837,
      "compression_loss": 0.0,
      "distillation_loss": 0.18467079102993011,
      "epoch": 6.07,
      "learning_rate": 2.153569298683017e-06,
      "loss": 0.1853,
      "step": 6394,
      "task_loss": 0.1906275451183319
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986006284532569,
      "compression_loss": 0.0,
      "distillation_loss": 0.09689827263355255,
      "epoch": 6.07,
      "learning_rate": 2.149244964034888e-06,
      "loss": 0.0951,
      "step": 6395,
      "task_loss": 0.07894426584243774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986026954714555,
      "compression_loss": 0.0,
      "distillation_loss": 0.043446026742458344,
      "epoch": 6.07,
      "learning_rate": 2.1449247803364687e-06,
      "loss": 0.0397,
      "step": 6396,
      "task_loss": 0.006183173507452011
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798604760453183,
      "compression_loss": 0.0,
      "distillation_loss": 0.05788339674472809,
      "epoch": 6.08,
      "learning_rate": 2.140608748372533e-06,
      "loss": 0.0535,
      "step": 6397,
      "task_loss": 0.013864126056432724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986068233994433,
      "compression_loss": 0.0,
      "distillation_loss": 0.06160301715135574,
      "epoch": 6.08,
      "learning_rate": 2.136296868927104e-06,
      "loss": 0.0612,
      "step": 6398,
      "task_loss": 0.057745207101106644
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79860888431124,
      "compression_loss": 0.0,
      "distillation_loss": 0.06886729598045349,
      "epoch": 6.08,
      "learning_rate": 2.1319891427834664e-06,
      "loss": 0.0634,
      "step": 6399,
      "task_loss": 0.014371348544955254
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986109431895767,
      "compression_loss": 0.0,
      "distillation_loss": 0.05757201835513115,
      "epoch": 6.08,
      "learning_rate": 2.127685570724136e-06,
      "loss": 0.0528,
      "step": 6400,
      "task_loss": 0.009750045835971832
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986130000354573,
      "compression_loss": 0.0,
      "distillation_loss": 0.1767224222421646,
      "epoch": 6.08,
      "learning_rate": 2.1233861535308706e-06,
      "loss": 0.1751,
      "step": 6401,
      "task_loss": 0.1601455807685852
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986150548498853,
      "compression_loss": 0.0,
      "distillation_loss": 0.08510977774858475,
      "epoch": 6.08,
      "learning_rate": 2.1190908919846875e-06,
      "loss": 0.08,
      "step": 6402,
      "task_loss": 0.03425342217087746
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986171076338643,
      "compression_loss": 0.0,
      "distillation_loss": 0.10737930238246918,
      "epoch": 6.08,
      "learning_rate": 2.1147997868658425e-06,
      "loss": 0.1043,
      "step": 6403,
      "task_loss": 0.0761018693447113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986191583883983,
      "compression_loss": 0.0,
      "distillation_loss": 0.265876829624176,
      "epoch": 6.08,
      "learning_rate": 2.110512838953832e-06,
      "loss": 0.2627,
      "step": 6404,
      "task_loss": 0.23371505737304688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986212071144907,
      "compression_loss": 0.0,
      "distillation_loss": 0.0776783674955368,
      "epoch": 6.08,
      "learning_rate": 2.1062300490273968e-06,
      "loss": 0.0801,
      "step": 6405,
      "task_loss": 0.10176847875118256
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986232538131453,
      "compression_loss": 0.0,
      "distillation_loss": 0.04853527992963791,
      "epoch": 6.08,
      "learning_rate": 2.1019514178645367e-06,
      "loss": 0.0567,
      "step": 6406,
      "task_loss": 0.13034237921237946
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986252984853657,
      "compression_loss": 0.0,
      "distillation_loss": 0.04432214796543121,
      "epoch": 6.08,
      "learning_rate": 2.0976769462424774e-06,
      "loss": 0.0406,
      "step": 6407,
      "task_loss": 0.007335290312767029
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986273411321556,
      "compression_loss": 0.0,
      "distillation_loss": 0.05146826058626175,
      "epoch": 6.09,
      "learning_rate": 2.0934066349376975e-06,
      "loss": 0.0476,
      "step": 6408,
      "task_loss": 0.012544374912977219
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986293817545188,
      "compression_loss": 0.0,
      "distillation_loss": 0.08119907975196838,
      "epoch": 6.09,
      "learning_rate": 2.0891404847259267e-06,
      "loss": 0.0824,
      "step": 6409,
      "task_loss": 0.09306506812572479
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986314203534588,
      "compression_loss": 0.0,
      "distillation_loss": 0.06213619187474251,
      "epoch": 6.09,
      "learning_rate": 2.08487849638212e-06,
      "loss": 0.0582,
      "step": 6410,
      "task_loss": 0.023112241178750992
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986334569299796,
      "compression_loss": 0.0,
      "distillation_loss": 0.06045207008719444,
      "epoch": 6.09,
      "learning_rate": 2.0806206706804998e-06,
      "loss": 0.0657,
      "step": 6411,
      "task_loss": 0.11293409764766693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986354914850844,
      "compression_loss": 0.0,
      "distillation_loss": 0.06079762801527977,
      "epoch": 6.09,
      "learning_rate": 2.0763670083945114e-06,
      "loss": 0.0659,
      "step": 6412,
      "task_loss": 0.11149650812149048
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986375240197773,
      "compression_loss": 0.0,
      "distillation_loss": 0.1584654152393341,
      "epoch": 6.09,
      "learning_rate": 2.0721175102968616e-06,
      "loss": 0.147,
      "step": 6413,
      "task_loss": 0.043657951056957245
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986395545350619,
      "compression_loss": 0.0,
      "distillation_loss": 0.06325273215770721,
      "epoch": 6.09,
      "learning_rate": 2.067872177159488e-06,
      "loss": 0.059,
      "step": 6414,
      "task_loss": 0.020560430362820625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986415830319417,
      "compression_loss": 0.0,
      "distillation_loss": 0.050118543207645416,
      "epoch": 6.09,
      "learning_rate": 2.0636310097535724e-06,
      "loss": 0.0472,
      "step": 6415,
      "task_loss": 0.020842991769313812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986436095114204,
      "compression_loss": 0.0,
      "distillation_loss": 0.06810344755649567,
      "epoch": 6.09,
      "learning_rate": 2.0593940088495495e-06,
      "loss": 0.0687,
      "step": 6416,
      "task_loss": 0.07414126396179199
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798645633974502,
      "compression_loss": 0.0,
      "distillation_loss": 0.10694337636232376,
      "epoch": 6.09,
      "learning_rate": 2.055161175217091e-06,
      "loss": 0.1054,
      "step": 6417,
      "task_loss": 0.09133277088403702
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986476564221897,
      "compression_loss": 0.0,
      "distillation_loss": 0.05698537826538086,
      "epoch": 6.09,
      "learning_rate": 2.0509325096251004e-06,
      "loss": 0.0552,
      "step": 6418,
      "task_loss": 0.03890334814786911
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986496768554876,
      "compression_loss": 0.0,
      "distillation_loss": 0.07449530810117722,
      "epoch": 6.1,
      "learning_rate": 2.046708012841744e-06,
      "loss": 0.072,
      "step": 6419,
      "task_loss": 0.049478065222501755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986516952753993,
      "compression_loss": 0.0,
      "distillation_loss": 0.07264258712530136,
      "epoch": 6.1,
      "learning_rate": 2.042487685634428e-06,
      "loss": 0.0676,
      "step": 6420,
      "task_loss": 0.021874142810702324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986537116829282,
      "compression_loss": 0.0,
      "distillation_loss": 0.09519660472869873,
      "epoch": 6.1,
      "learning_rate": 2.038271528769786e-06,
      "loss": 0.1082,
      "step": 6421,
      "task_loss": 0.2249455749988556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986557260790783,
      "compression_loss": 0.0,
      "distillation_loss": 0.03500593453645706,
      "epoch": 6.1,
      "learning_rate": 2.034059543013703e-06,
      "loss": 0.0322,
      "step": 6422,
      "task_loss": 0.006475191563367844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986577384648531,
      "compression_loss": 0.0,
      "distillation_loss": 0.0777549147605896,
      "epoch": 6.1,
      "learning_rate": 2.029851729131313e-06,
      "loss": 0.081,
      "step": 6423,
      "task_loss": 0.10977138578891754
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986597488412565,
      "compression_loss": 0.0,
      "distillation_loss": 0.12887020409107208,
      "epoch": 6.1,
      "learning_rate": 2.0256480878869834e-06,
      "loss": 0.1275,
      "step": 6424,
      "task_loss": 0.11558264493942261
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986617572092919,
      "compression_loss": 0.0,
      "distillation_loss": 0.1287572830915451,
      "epoch": 6.1,
      "learning_rate": 2.0214486200443168e-06,
      "loss": 0.1278,
      "step": 6425,
      "task_loss": 0.11913389712572098
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986637635699632,
      "compression_loss": 0.0,
      "distillation_loss": 0.06908180564641953,
      "epoch": 6.1,
      "learning_rate": 2.017253326366181e-06,
      "loss": 0.065,
      "step": 6426,
      "task_loss": 0.027965903282165527
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986657679242739,
      "compression_loss": 0.0,
      "distillation_loss": 0.09820552915334702,
      "epoch": 6.1,
      "learning_rate": 2.0130622076146576e-06,
      "loss": 0.0986,
      "step": 6427,
      "task_loss": 0.10234321653842926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986677702732278,
      "compression_loss": 0.0,
      "distillation_loss": 0.0638742744922638,
      "epoch": 6.1,
      "learning_rate": 2.0088752645510995e-06,
      "loss": 0.0609,
      "step": 6428,
      "task_loss": 0.03451484069228172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986697706178286,
      "compression_loss": 0.0,
      "distillation_loss": 0.18671873211860657,
      "epoch": 6.11,
      "learning_rate": 2.004692497936067e-06,
      "loss": 0.194,
      "step": 6429,
      "task_loss": 0.259968101978302
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986717689590799,
      "compression_loss": 0.0,
      "distillation_loss": 0.04738321155309677,
      "epoch": 6.11,
      "learning_rate": 2.0005139085293945e-06,
      "loss": 0.0452,
      "step": 6430,
      "task_loss": 0.025601202622056007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986737652979855,
      "compression_loss": 0.0,
      "distillation_loss": 0.05693122744560242,
      "epoch": 6.11,
      "learning_rate": 1.996339497090138e-06,
      "loss": 0.0529,
      "step": 6431,
      "task_loss": 0.01685180887579918
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986757596355489,
      "compression_loss": 0.0,
      "distillation_loss": 0.0986439436674118,
      "epoch": 6.11,
      "learning_rate": 1.9921692643765947e-06,
      "loss": 0.0996,
      "step": 6432,
      "task_loss": 0.10801742225885391
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798677751972774,
      "compression_loss": 0.0,
      "distillation_loss": 0.11213336884975433,
      "epoch": 6.11,
      "learning_rate": 1.9880032111463166e-06,
      "loss": 0.1301,
      "step": 6433,
      "task_loss": 0.29207485914230347
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986797423106643,
      "compression_loss": 0.0,
      "distillation_loss": 0.035904861986637115,
      "epoch": 6.11,
      "learning_rate": 1.9838413381560813e-06,
      "loss": 0.0345,
      "step": 6434,
      "task_loss": 0.021603219211101532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986817306502235,
      "compression_loss": 0.0,
      "distillation_loss": 0.05209188535809517,
      "epoch": 6.11,
      "learning_rate": 1.979683646161909e-06,
      "loss": 0.0516,
      "step": 6435,
      "task_loss": 0.04718216508626938
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986837169924554,
      "compression_loss": 0.0,
      "distillation_loss": 0.05253394693136215,
      "epoch": 6.11,
      "learning_rate": 1.975530135919068e-06,
      "loss": 0.0616,
      "step": 6436,
      "task_loss": 0.14322909712791443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986857013383637,
      "compression_loss": 0.0,
      "distillation_loss": 0.0527234822511673,
      "epoch": 6.11,
      "learning_rate": 1.9713808081820716e-06,
      "loss": 0.0584,
      "step": 6437,
      "task_loss": 0.1097893938422203
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986876836889519,
      "compression_loss": 0.0,
      "distillation_loss": 0.036524612456560135,
      "epoch": 6.11,
      "learning_rate": 1.9672356637046584e-06,
      "loss": 0.0335,
      "step": 6438,
      "task_loss": 0.00587356835603714
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986896640452238,
      "compression_loss": 0.0,
      "distillation_loss": 0.0352536141872406,
      "epoch": 6.11,
      "learning_rate": 1.9630947032398067e-06,
      "loss": 0.0478,
      "step": 6439,
      "task_loss": 0.16093014180660248
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798691642408183,
      "compression_loss": 0.0,
      "distillation_loss": 0.04243883490562439,
      "epoch": 6.12,
      "learning_rate": 1.9589579275397562e-06,
      "loss": 0.044,
      "step": 6440,
      "task_loss": 0.058299075812101364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986936187788334,
      "compression_loss": 0.0,
      "distillation_loss": 0.1163100153207779,
      "epoch": 6.12,
      "learning_rate": 1.9548253373559646e-06,
      "loss": 0.1117,
      "step": 6441,
      "task_loss": 0.07047397643327713
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986955931581784,
      "compression_loss": 0.0,
      "distillation_loss": 0.0603814423084259,
      "epoch": 6.12,
      "learning_rate": 1.9506969334391332e-06,
      "loss": 0.0554,
      "step": 6442,
      "task_loss": 0.010571654886007309
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986975655472218,
      "compression_loss": 0.0,
      "distillation_loss": 0.03320575878024101,
      "epoch": 6.12,
      "learning_rate": 1.9465727165392134e-06,
      "loss": 0.0305,
      "step": 6443,
      "task_loss": 0.006414549425244331
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7986995359469674,
      "compression_loss": 0.0,
      "distillation_loss": 0.025436367839574814,
      "epoch": 6.12,
      "learning_rate": 1.942452687405383e-06,
      "loss": 0.0307,
      "step": 6444,
      "task_loss": 0.07796012610197067
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987015043584187,
      "compression_loss": 0.0,
      "distillation_loss": 0.04995856434106827,
      "epoch": 6.12,
      "learning_rate": 1.9383368467860734e-06,
      "loss": 0.0519,
      "step": 6445,
      "task_loss": 0.06982216984033585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987034707825795,
      "compression_loss": 0.0,
      "distillation_loss": 0.028508171439170837,
      "epoch": 6.12,
      "learning_rate": 1.9342251954289346e-06,
      "loss": 0.0264,
      "step": 6446,
      "task_loss": 0.00761030986905098
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987054352204533,
      "compression_loss": 0.0,
      "distillation_loss": 0.06900826841592789,
      "epoch": 6.12,
      "learning_rate": 1.930117734080883e-06,
      "loss": 0.0651,
      "step": 6447,
      "task_loss": 0.029789365828037262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987073976730441,
      "compression_loss": 0.0,
      "distillation_loss": 0.07841695845127106,
      "epoch": 6.12,
      "learning_rate": 1.926014463488049e-06,
      "loss": 0.0737,
      "step": 6448,
      "task_loss": 0.031337086111307144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987093581413554,
      "compression_loss": 0.0,
      "distillation_loss": 0.03884424641728401,
      "epoch": 6.12,
      "learning_rate": 1.921915384395809e-06,
      "loss": 0.0368,
      "step": 6449,
      "task_loss": 0.018084675073623657
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987113166263907,
      "compression_loss": 0.0,
      "distillation_loss": 0.03938658908009529,
      "epoch": 6.13,
      "learning_rate": 1.917820497548789e-06,
      "loss": 0.048,
      "step": 6450,
      "task_loss": 0.12579363584518433
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987132731291541,
      "compression_loss": 0.0,
      "distillation_loss": 0.0376831591129303,
      "epoch": 6.13,
      "learning_rate": 1.913729803690839e-06,
      "loss": 0.0408,
      "step": 6451,
      "task_loss": 0.06835294514894485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987152276506488,
      "compression_loss": 0.0,
      "distillation_loss": 0.07034672051668167,
      "epoch": 6.13,
      "learning_rate": 1.9096433035650565e-06,
      "loss": 0.0667,
      "step": 6452,
      "task_loss": 0.03402462229132652
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987171801918789,
      "compression_loss": 0.0,
      "distillation_loss": 0.06073524430394173,
      "epoch": 6.13,
      "learning_rate": 1.9055609979137634e-06,
      "loss": 0.0568,
      "step": 6453,
      "task_loss": 0.020950062200427055
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987191307538479,
      "compression_loss": 0.0,
      "distillation_loss": 0.08777901530265808,
      "epoch": 6.13,
      "learning_rate": 1.9014828874785478e-06,
      "loss": 0.0853,
      "step": 6454,
      "task_loss": 0.0634586289525032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987210793375594,
      "compression_loss": 0.0,
      "distillation_loss": 0.054110340774059296,
      "epoch": 6.13,
      "learning_rate": 1.8974089730002087e-06,
      "loss": 0.0514,
      "step": 6455,
      "task_loss": 0.026968229562044144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987230259440173,
      "compression_loss": 0.0,
      "distillation_loss": 0.03255057334899902,
      "epoch": 6.13,
      "learning_rate": 1.89333925521879e-06,
      "loss": 0.0306,
      "step": 6456,
      "task_loss": 0.012770043686032295
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987249705742251,
      "compression_loss": 0.0,
      "distillation_loss": 0.05460744723677635,
      "epoch": 6.13,
      "learning_rate": 1.8892737348735812e-06,
      "loss": 0.0586,
      "step": 6457,
      "task_loss": 0.09443493187427521
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987269132291865,
      "compression_loss": 0.0,
      "distillation_loss": 0.10842868685722351,
      "epoch": 6.13,
      "learning_rate": 1.8852124127031022e-06,
      "loss": 0.1172,
      "step": 6458,
      "task_loss": 0.19569729268550873
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987288539099053,
      "compression_loss": 0.0,
      "distillation_loss": 0.04304853081703186,
      "epoch": 6.13,
      "learning_rate": 1.8811552894451107e-06,
      "loss": 0.042,
      "step": 6459,
      "task_loss": 0.03260982409119606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987307926173851,
      "compression_loss": 0.0,
      "distillation_loss": 0.07537424564361572,
      "epoch": 6.13,
      "learning_rate": 1.877102365836597e-06,
      "loss": 0.0735,
      "step": 6460,
      "task_loss": 0.056225549429655075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987327293526295,
      "compression_loss": 0.0,
      "distillation_loss": 0.02507023513317108,
      "epoch": 6.14,
      "learning_rate": 1.8730536426138034e-06,
      "loss": 0.023,
      "step": 6461,
      "task_loss": 0.004859650507569313
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987346641166423,
      "compression_loss": 0.0,
      "distillation_loss": 0.06949448585510254,
      "epoch": 6.14,
      "learning_rate": 1.8690091205121941e-06,
      "loss": 0.0698,
      "step": 6462,
      "task_loss": 0.07217264175415039
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987365969104272,
      "compression_loss": 0.0,
      "distillation_loss": 0.04962468147277832,
      "epoch": 6.14,
      "learning_rate": 1.8649688002664756e-06,
      "loss": 0.048,
      "step": 6463,
      "task_loss": 0.033658094704151154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987385277349879,
      "compression_loss": 0.0,
      "distillation_loss": 0.02775590494275093,
      "epoch": 6.14,
      "learning_rate": 1.8609326826106e-06,
      "loss": 0.0255,
      "step": 6464,
      "task_loss": 0.00518425740301609
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987404565913279,
      "compression_loss": 0.0,
      "distillation_loss": 0.029499584808945656,
      "epoch": 6.14,
      "learning_rate": 1.8569007682777417e-06,
      "loss": 0.0272,
      "step": 6465,
      "task_loss": 0.006057474762201309
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798742383480451,
      "compression_loss": 0.0,
      "distillation_loss": 0.04246009886264801,
      "epoch": 6.14,
      "learning_rate": 1.8528730580003178e-06,
      "loss": 0.0466,
      "step": 6466,
      "task_loss": 0.08337102830410004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987443084033609,
      "compression_loss": 0.0,
      "distillation_loss": 0.08518815040588379,
      "epoch": 6.14,
      "learning_rate": 1.8488495525099735e-06,
      "loss": 0.0868,
      "step": 6467,
      "task_loss": 0.10114063322544098
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987462313610613,
      "compression_loss": 0.0,
      "distillation_loss": 0.04680711403489113,
      "epoch": 6.14,
      "learning_rate": 1.8448302525376132e-06,
      "loss": 0.0431,
      "step": 6468,
      "task_loss": 0.009998075664043427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987481523545558,
      "compression_loss": 0.0,
      "distillation_loss": 0.07749959826469421,
      "epoch": 6.14,
      "learning_rate": 1.8408151588133498e-06,
      "loss": 0.0829,
      "step": 6469,
      "task_loss": 0.13170292973518372
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987500713848481,
      "compression_loss": 0.0,
      "distillation_loss": 0.10043874382972717,
      "epoch": 6.14,
      "learning_rate": 1.8368042720665446e-06,
      "loss": 0.1057,
      "step": 6470,
      "task_loss": 0.15264543890953064
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987519884529419,
      "compression_loss": 0.0,
      "distillation_loss": 0.10912171006202698,
      "epoch": 6.15,
      "learning_rate": 1.8327975930258035e-06,
      "loss": 0.1048,
      "step": 6471,
      "task_loss": 0.0659012421965599
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798753903559841,
      "compression_loss": 0.0,
      "distillation_loss": 0.01361011527478695,
      "epoch": 6.15,
      "learning_rate": 1.8287951224189553e-06,
      "loss": 0.0198,
      "step": 6472,
      "task_loss": 0.07584784179925919
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987558167065488,
      "compression_loss": 0.0,
      "distillation_loss": 0.14504070580005646,
      "epoch": 6.15,
      "learning_rate": 1.8247968609730686e-06,
      "loss": 0.1404,
      "step": 6473,
      "task_loss": 0.09844270348548889
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987577278940693,
      "compression_loss": 0.0,
      "distillation_loss": 0.07923628389835358,
      "epoch": 6.15,
      "learning_rate": 1.8208028094144375e-06,
      "loss": 0.0724,
      "step": 6474,
      "task_loss": 0.011291364207863808
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798759637123406,
      "compression_loss": 0.0,
      "distillation_loss": 0.023012623190879822,
      "epoch": 6.15,
      "learning_rate": 1.8168129684686148e-06,
      "loss": 0.0308,
      "step": 6475,
      "task_loss": 0.10104191303253174
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987615443955627,
      "compression_loss": 0.0,
      "distillation_loss": 0.037004388868808746,
      "epoch": 6.15,
      "learning_rate": 1.8128273388603679e-06,
      "loss": 0.0342,
      "step": 6476,
      "task_loss": 0.008559944108128548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987634497115429,
      "compression_loss": 0.0,
      "distillation_loss": 0.02188132330775261,
      "epoch": 6.15,
      "learning_rate": 1.808845921313701e-06,
      "loss": 0.0206,
      "step": 6477,
      "task_loss": 0.008831024169921875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987653530723503,
      "compression_loss": 0.0,
      "distillation_loss": 0.02209128439426422,
      "epoch": 6.15,
      "learning_rate": 1.8048687165518662e-06,
      "loss": 0.0204,
      "step": 6478,
      "task_loss": 0.004884377121925354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987672544789888,
      "compression_loss": 0.0,
      "distillation_loss": 0.026295151561498642,
      "epoch": 6.15,
      "learning_rate": 1.800895725297333e-06,
      "loss": 0.0385,
      "step": 6479,
      "task_loss": 0.14823564887046814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798769153932462,
      "compression_loss": 0.0,
      "distillation_loss": 0.02808169648051262,
      "epoch": 6.15,
      "learning_rate": 1.7969269482718265e-06,
      "loss": 0.0283,
      "step": 6480,
      "task_loss": 0.0304415300488472
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987710514337734,
      "compression_loss": 0.0,
      "distillation_loss": 0.09501353651285172,
      "epoch": 6.15,
      "learning_rate": 1.7929623861962785e-06,
      "loss": 0.1057,
      "step": 6481,
      "task_loss": 0.2016463577747345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987729469839269,
      "compression_loss": 0.0,
      "distillation_loss": 0.05659741908311844,
      "epoch": 6.16,
      "learning_rate": 1.789002039790888e-06,
      "loss": 0.0551,
      "step": 6482,
      "task_loss": 0.041581884026527405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798774840583926,
      "compression_loss": 0.0,
      "distillation_loss": 0.02717326581478119,
      "epoch": 6.16,
      "learning_rate": 1.78504590977506e-06,
      "loss": 0.0309,
      "step": 6483,
      "task_loss": 0.06475157290697098
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987767322347745,
      "compression_loss": 0.0,
      "distillation_loss": 0.10455787181854248,
      "epoch": 6.16,
      "learning_rate": 1.7810939968674418e-06,
      "loss": 0.1024,
      "step": 6484,
      "task_loss": 0.08274058252573013
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987786219374761,
      "compression_loss": 0.0,
      "distillation_loss": 0.06307333707809448,
      "epoch": 6.16,
      "learning_rate": 1.7771463017859287e-06,
      "loss": 0.0622,
      "step": 6485,
      "task_loss": 0.05453290790319443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987805096930344,
      "compression_loss": 0.0,
      "distillation_loss": 0.05282985419034958,
      "epoch": 6.16,
      "learning_rate": 1.77320282524763e-06,
      "loss": 0.0512,
      "step": 6486,
      "task_loss": 0.03684225305914879
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987823955024531,
      "compression_loss": 0.0,
      "distillation_loss": 0.04597539082169533,
      "epoch": 6.16,
      "learning_rate": 1.7692635679688986e-06,
      "loss": 0.0523,
      "step": 6487,
      "task_loss": 0.10876553505659103
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987842793667359,
      "compression_loss": 0.0,
      "distillation_loss": 0.08502782136201859,
      "epoch": 6.16,
      "learning_rate": 1.7653285306653194e-06,
      "loss": 0.0824,
      "step": 6488,
      "task_loss": 0.05903245136141777
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987861612868866,
      "compression_loss": 0.0,
      "distillation_loss": 0.16358347237110138,
      "epoch": 6.16,
      "learning_rate": 1.7613977140517158e-06,
      "loss": 0.1704,
      "step": 6489,
      "task_loss": 0.23221097886562347
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987880412639087,
      "compression_loss": 0.0,
      "distillation_loss": 0.05940496176481247,
      "epoch": 6.16,
      "learning_rate": 1.7574711188421356e-06,
      "loss": 0.0593,
      "step": 6490,
      "task_loss": 0.05786725506186485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987899192988059,
      "compression_loss": 0.0,
      "distillation_loss": 0.05588078126311302,
      "epoch": 6.16,
      "learning_rate": 1.7535487457498583e-06,
      "loss": 0.0596,
      "step": 6491,
      "task_loss": 0.09317133575677872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798791795392582,
      "compression_loss": 0.0,
      "distillation_loss": 0.0370253287255764,
      "epoch": 6.17,
      "learning_rate": 1.7496305954874142e-06,
      "loss": 0.0443,
      "step": 6492,
      "task_loss": 0.10931254923343658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987936695462405,
      "compression_loss": 0.0,
      "distillation_loss": 0.03799361735582352,
      "epoch": 6.17,
      "learning_rate": 1.7457166687665449e-06,
      "loss": 0.045,
      "step": 6493,
      "task_loss": 0.1080305278301239
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987955417607854,
      "compression_loss": 0.0,
      "distillation_loss": 0.0513494536280632,
      "epoch": 6.17,
      "learning_rate": 1.7418069662982344e-06,
      "loss": 0.0493,
      "step": 6494,
      "task_loss": 0.03114178031682968
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987974120372201,
      "compression_loss": 0.0,
      "distillation_loss": 0.0808219462633133,
      "epoch": 6.17,
      "learning_rate": 1.7379014887927064e-06,
      "loss": 0.0772,
      "step": 6495,
      "task_loss": 0.044266168028116226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7987992803765482,
      "compression_loss": 0.0,
      "distillation_loss": 0.19209139049053192,
      "epoch": 6.17,
      "learning_rate": 1.734000236959399e-06,
      "loss": 0.1894,
      "step": 6496,
      "task_loss": 0.16527491807937622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988011467797738,
      "compression_loss": 0.0,
      "distillation_loss": 0.033167578279972076,
      "epoch": 6.17,
      "learning_rate": 1.7301032115070003e-06,
      "loss": 0.0421,
      "step": 6497,
      "task_loss": 0.12249952554702759
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988030112479001,
      "compression_loss": 0.0,
      "distillation_loss": 0.061443835496902466,
      "epoch": 6.17,
      "learning_rate": 1.7262104131434226e-06,
      "loss": 0.0669,
      "step": 6498,
      "task_loss": 0.11603046953678131
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988048737819311,
      "compression_loss": 0.0,
      "distillation_loss": 0.020476579666137695,
      "epoch": 6.17,
      "learning_rate": 1.722321842575811e-06,
      "loss": 0.0188,
      "step": 6499,
      "task_loss": 0.0034268908202648163
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988067343828704,
      "compression_loss": 0.0,
      "distillation_loss": 0.0780348926782608,
      "epoch": 6.17,
      "learning_rate": 1.7184375005105474e-06,
      "loss": 0.0799,
      "step": 6500,
      "task_loss": 0.09668619930744171
    },
    {
      "epoch": 6.17,
      "eval_accuracy": 0.8876146788990825,
      "eval_loss": 0.4523507356643677,
      "eval_runtime": 18.245,
      "eval_samples_per_second": 47.794,
      "eval_steps_per_second": 5.974,
      "step": 6500
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988085930517217,
      "compression_loss": 0.0,
      "distillation_loss": 0.019627831876277924,
      "epoch": 6.17,
      "learning_rate": 1.714557387653229e-06,
      "loss": 0.0234,
      "step": 6501,
      "task_loss": 0.0573212131857872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988104497894886,
      "compression_loss": 0.0,
      "distillation_loss": 0.04171549528837204,
      "epoch": 6.17,
      "learning_rate": 1.710681504708711e-06,
      "loss": 0.0404,
      "step": 6502,
      "task_loss": 0.028218252584338188
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988123045971749,
      "compression_loss": 0.0,
      "distillation_loss": 0.03882744908332825,
      "epoch": 6.18,
      "learning_rate": 1.7068098523810611e-06,
      "loss": 0.0428,
      "step": 6503,
      "task_loss": 0.07813085615634918
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988141574757841,
      "compression_loss": 0.0,
      "distillation_loss": 0.028121117502450943,
      "epoch": 6.18,
      "learning_rate": 1.7029424313735776e-06,
      "loss": 0.0338,
      "step": 6504,
      "task_loss": 0.08514552563428879
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988160084263201,
      "compression_loss": 0.0,
      "distillation_loss": 0.0896795317530632,
      "epoch": 6.18,
      "learning_rate": 1.6990792423888013e-06,
      "loss": 0.1013,
      "step": 6505,
      "task_loss": 0.20579922199249268
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988178574497864,
      "compression_loss": 0.0,
      "distillation_loss": 0.09015851467847824,
      "epoch": 6.18,
      "learning_rate": 1.6952202861285044e-06,
      "loss": 0.0904,
      "step": 6506,
      "task_loss": 0.09299979358911514
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988197045471869,
      "compression_loss": 0.0,
      "distillation_loss": 0.21845272183418274,
      "epoch": 6.18,
      "learning_rate": 1.6913655632936787e-06,
      "loss": 0.2116,
      "step": 6507,
      "task_loss": 0.1496710181236267
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798821549719525,
      "compression_loss": 0.0,
      "distillation_loss": 0.05457516014575958,
      "epoch": 6.18,
      "learning_rate": 1.6875150745845503e-06,
      "loss": 0.0568,
      "step": 6508,
      "task_loss": 0.07647211849689484
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988233929678046,
      "compression_loss": 0.0,
      "distillation_loss": 0.059655383229255676,
      "epoch": 6.18,
      "learning_rate": 1.6836688207005846e-06,
      "loss": 0.0653,
      "step": 6509,
      "task_loss": 0.11646562069654465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988252342930292,
      "compression_loss": 0.0,
      "distillation_loss": 0.06423240900039673,
      "epoch": 6.18,
      "learning_rate": 1.6798268023404727e-06,
      "loss": 0.0602,
      "step": 6510,
      "task_loss": 0.023659339174628258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988270736962028,
      "compression_loss": 0.0,
      "distillation_loss": 0.06197018176317215,
      "epoch": 6.18,
      "learning_rate": 1.6759890202021289e-06,
      "loss": 0.0629,
      "step": 6511,
      "task_loss": 0.07137607038021088
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988289111783287,
      "compression_loss": 0.0,
      "distillation_loss": 0.026029150933027267,
      "epoch": 6.18,
      "learning_rate": 1.6721554749827116e-06,
      "loss": 0.0309,
      "step": 6512,
      "task_loss": 0.07451260089874268
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988307467404108,
      "compression_loss": 0.0,
      "distillation_loss": 0.04224439710378647,
      "epoch": 6.19,
      "learning_rate": 1.6683261673786033e-06,
      "loss": 0.052,
      "step": 6513,
      "task_loss": 0.14017590880393982
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988325803834527,
      "compression_loss": 0.0,
      "distillation_loss": 0.06829048693180084,
      "epoch": 6.19,
      "learning_rate": 1.664501098085408e-06,
      "loss": 0.0639,
      "step": 6514,
      "task_loss": 0.024488359689712524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988344121084583,
      "compression_loss": 0.0,
      "distillation_loss": 0.08032073080539703,
      "epoch": 6.19,
      "learning_rate": 1.6606802677979732e-06,
      "loss": 0.0757,
      "step": 6515,
      "task_loss": 0.03441820666193962
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988362419164309,
      "compression_loss": 0.0,
      "distillation_loss": 0.025070223957300186,
      "epoch": 6.19,
      "learning_rate": 1.656863677210374e-06,
      "loss": 0.0523,
      "step": 6516,
      "task_loss": 0.2975619435310364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988380698083745,
      "compression_loss": 0.0,
      "distillation_loss": 0.035705260932445526,
      "epoch": 6.19,
      "learning_rate": 1.6530513270159116e-06,
      "loss": 0.0358,
      "step": 6517,
      "task_loss": 0.03673187643289566
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988398957852926,
      "compression_loss": 0.0,
      "distillation_loss": 0.09474039822816849,
      "epoch": 6.19,
      "learning_rate": 1.6492432179071094e-06,
      "loss": 0.0979,
      "step": 6518,
      "task_loss": 0.12658804655075073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798841719848189,
      "compression_loss": 0.0,
      "distillation_loss": 0.057927001267671585,
      "epoch": 6.19,
      "learning_rate": 1.645439350575742e-06,
      "loss": 0.0682,
      "step": 6519,
      "task_loss": 0.16106078028678894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988435419980673,
      "compression_loss": 0.0,
      "distillation_loss": 0.05671095848083496,
      "epoch": 6.19,
      "learning_rate": 1.6416397257127902e-06,
      "loss": 0.0594,
      "step": 6520,
      "task_loss": 0.08323853462934494
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988453622359312,
      "compression_loss": 0.0,
      "distillation_loss": 0.06431238353252411,
      "epoch": 6.19,
      "learning_rate": 1.637844344008474e-06,
      "loss": 0.0647,
      "step": 6521,
      "task_loss": 0.06814312934875488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988471805627844,
      "compression_loss": 0.0,
      "distillation_loss": 0.03375518321990967,
      "epoch": 6.19,
      "learning_rate": 1.6340532061522474e-06,
      "loss": 0.0327,
      "step": 6522,
      "task_loss": 0.022706888616085052
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988489969796306,
      "compression_loss": 0.0,
      "distillation_loss": 0.052014727145433426,
      "epoch": 6.19,
      "learning_rate": 1.6302663128327927e-06,
      "loss": 0.0496,
      "step": 6523,
      "task_loss": 0.028193766251206398
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988508114874735,
      "compression_loss": 0.0,
      "distillation_loss": 0.10802538692951202,
      "epoch": 6.2,
      "learning_rate": 1.62648366473801e-06,
      "loss": 0.104,
      "step": 6524,
      "task_loss": 0.06763241440057755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988526240873166,
      "compression_loss": 0.0,
      "distillation_loss": 0.05575351417064667,
      "epoch": 6.2,
      "learning_rate": 1.6227052625550327e-06,
      "loss": 0.0649,
      "step": 6525,
      "task_loss": 0.14689795672893524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988544347801638,
      "compression_loss": 0.0,
      "distillation_loss": 0.08440142869949341,
      "epoch": 6.2,
      "learning_rate": 1.6189311069702367e-06,
      "loss": 0.0887,
      "step": 6526,
      "task_loss": 0.12698373198509216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988562435670188,
      "compression_loss": 0.0,
      "distillation_loss": 0.04269465059041977,
      "epoch": 6.2,
      "learning_rate": 1.61516119866921e-06,
      "loss": 0.047,
      "step": 6527,
      "task_loss": 0.08578742295503616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798858050448885,
      "compression_loss": 0.0,
      "distillation_loss": 0.024140551686286926,
      "epoch": 6.2,
      "learning_rate": 1.6113955383367685e-06,
      "loss": 0.024,
      "step": 6528,
      "task_loss": 0.022829843685030937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988598554267664,
      "compression_loss": 0.0,
      "distillation_loss": 0.033472198992967606,
      "epoch": 6.2,
      "learning_rate": 1.6076341266569734e-06,
      "loss": 0.0353,
      "step": 6529,
      "task_loss": 0.05162709578871727
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988616585016666,
      "compression_loss": 0.0,
      "distillation_loss": 0.04639057815074921,
      "epoch": 6.2,
      "learning_rate": 1.6038769643130973e-06,
      "loss": 0.0581,
      "step": 6530,
      "task_loss": 0.1631542444229126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988634596745892,
      "compression_loss": 0.0,
      "distillation_loss": 0.041402582079172134,
      "epoch": 6.2,
      "learning_rate": 1.600124051987645e-06,
      "loss": 0.0542,
      "step": 6531,
      "task_loss": 0.16910916566848755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988652589465378,
      "compression_loss": 0.0,
      "distillation_loss": 0.045691799372434616,
      "epoch": 6.2,
      "learning_rate": 1.5963753903623535e-06,
      "loss": 0.0528,
      "step": 6532,
      "task_loss": 0.11651049554347992
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988670563185163,
      "compression_loss": 0.0,
      "distillation_loss": 0.03516625612974167,
      "epoch": 6.2,
      "learning_rate": 1.592630980118187e-06,
      "loss": 0.034,
      "step": 6533,
      "task_loss": 0.02300405688583851
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988688517915282,
      "compression_loss": 0.0,
      "distillation_loss": 0.039826877415180206,
      "epoch": 6.21,
      "learning_rate": 1.5888908219353349e-06,
      "loss": 0.039,
      "step": 6534,
      "task_loss": 0.03174047917127609
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988706453665774,
      "compression_loss": 0.0,
      "distillation_loss": 0.036477230489254,
      "epoch": 6.21,
      "learning_rate": 1.5851549164932116e-06,
      "loss": 0.034,
      "step": 6535,
      "task_loss": 0.011402864009141922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988724370446674,
      "compression_loss": 0.0,
      "distillation_loss": 0.05315593630075455,
      "epoch": 6.21,
      "learning_rate": 1.5814232644704691e-06,
      "loss": 0.0655,
      "step": 6536,
      "task_loss": 0.17702096700668335
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988742268268018,
      "compression_loss": 0.0,
      "distillation_loss": 0.1766928732395172,
      "epoch": 6.21,
      "learning_rate": 1.5776958665449731e-06,
      "loss": 0.1711,
      "step": 6537,
      "task_loss": 0.12121965736150742
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988760147139845,
      "compression_loss": 0.0,
      "distillation_loss": 0.04658925160765648,
      "epoch": 6.21,
      "learning_rate": 1.5739727233938239e-06,
      "loss": 0.0449,
      "step": 6538,
      "task_loss": 0.029311183840036392
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988778007072191,
      "compression_loss": 0.0,
      "distillation_loss": 0.038374051451683044,
      "epoch": 6.21,
      "learning_rate": 1.5702538356933555e-06,
      "loss": 0.0453,
      "step": 6539,
      "task_loss": 0.1074916422367096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988795848075092,
      "compression_loss": 0.0,
      "distillation_loss": 0.026442725211381912,
      "epoch": 6.21,
      "learning_rate": 1.5665392041191107e-06,
      "loss": 0.0276,
      "step": 6540,
      "task_loss": 0.03780139237642288
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988813670158587,
      "compression_loss": 0.0,
      "distillation_loss": 0.025853261351585388,
      "epoch": 6.21,
      "learning_rate": 1.5628288293458804e-06,
      "loss": 0.0292,
      "step": 6541,
      "task_loss": 0.058958619832992554
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798883147333271,
      "compression_loss": 0.0,
      "distillation_loss": 0.06454972922801971,
      "epoch": 6.21,
      "learning_rate": 1.5591227120476643e-06,
      "loss": 0.0724,
      "step": 6542,
      "task_loss": 0.1430365890264511
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79888492576075,
      "compression_loss": 0.0,
      "distillation_loss": 0.025234917178750038,
      "epoch": 6.21,
      "learning_rate": 1.5554208528977044e-06,
      "loss": 0.0235,
      "step": 6543,
      "task_loss": 0.007690908387303352
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988867022992993,
      "compression_loss": 0.0,
      "distillation_loss": 0.06024536117911339,
      "epoch": 6.21,
      "learning_rate": 1.5517232525684571e-06,
      "loss": 0.0693,
      "step": 6544,
      "task_loss": 0.15101151168346405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988884769499225,
      "compression_loss": 0.0,
      "distillation_loss": 0.04414193704724312,
      "epoch": 6.22,
      "learning_rate": 1.548029911731605e-06,
      "loss": 0.0416,
      "step": 6545,
      "task_loss": 0.019007055088877678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988902497136234,
      "compression_loss": 0.0,
      "distillation_loss": 0.028746753931045532,
      "epoch": 6.22,
      "learning_rate": 1.5443408310580692e-06,
      "loss": 0.0323,
      "step": 6546,
      "task_loss": 0.06452102959156036
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988920205914057,
      "compression_loss": 0.0,
      "distillation_loss": 0.02359924092888832,
      "epoch": 6.22,
      "learning_rate": 1.5406560112179864e-06,
      "loss": 0.0291,
      "step": 6547,
      "task_loss": 0.07876047492027283
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988937895842729,
      "compression_loss": 0.0,
      "distillation_loss": 0.020535187795758247,
      "epoch": 6.22,
      "learning_rate": 1.5369754528807152e-06,
      "loss": 0.019,
      "step": 6548,
      "task_loss": 0.005380989983677864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798895556693229,
      "compression_loss": 0.0,
      "distillation_loss": 0.04782063886523247,
      "epoch": 6.22,
      "learning_rate": 1.5332991567148515e-06,
      "loss": 0.058,
      "step": 6549,
      "task_loss": 0.1497877985239029
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988973219192774,
      "compression_loss": 0.0,
      "distillation_loss": 0.024879930540919304,
      "epoch": 6.22,
      "learning_rate": 1.5296271233882165e-06,
      "loss": 0.0321,
      "step": 6550,
      "task_loss": 0.09727062284946442
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7988990852634219,
      "compression_loss": 0.0,
      "distillation_loss": 0.040715500712394714,
      "epoch": 6.22,
      "learning_rate": 1.5259593535678491e-06,
      "loss": 0.0399,
      "step": 6551,
      "task_loss": 0.03258955478668213
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989008467266661,
      "compression_loss": 0.0,
      "distillation_loss": 0.03168099373579025,
      "epoch": 6.22,
      "learning_rate": 1.522295847920019e-06,
      "loss": 0.0492,
      "step": 6552,
      "task_loss": 0.20730939507484436
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989026063100138,
      "compression_loss": 0.0,
      "distillation_loss": 0.08038985729217529,
      "epoch": 6.22,
      "learning_rate": 1.5186366071102133e-06,
      "loss": 0.0947,
      "step": 6553,
      "task_loss": 0.22315886616706848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989043640144685,
      "compression_loss": 0.0,
      "distillation_loss": 0.022372357547283173,
      "epoch": 6.22,
      "learning_rate": 1.5149816318031584e-06,
      "loss": 0.0208,
      "step": 6554,
      "task_loss": 0.00664399191737175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989061198410341,
      "compression_loss": 0.0,
      "distillation_loss": 0.06297887861728668,
      "epoch": 6.23,
      "learning_rate": 1.5113309226627986e-06,
      "loss": 0.0727,
      "step": 6555,
      "task_loss": 0.16039499640464783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989078737907143,
      "compression_loss": 0.0,
      "distillation_loss": 0.06179758906364441,
      "epoch": 6.23,
      "learning_rate": 1.5076844803522922e-06,
      "loss": 0.0664,
      "step": 6556,
      "task_loss": 0.10753493010997772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989096258645125,
      "compression_loss": 0.0,
      "distillation_loss": 0.025135168805718422,
      "epoch": 6.23,
      "learning_rate": 1.5040423055340396e-06,
      "loss": 0.0317,
      "step": 6557,
      "task_loss": 0.09094193577766418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989113760634325,
      "compression_loss": 0.0,
      "distillation_loss": 0.045239001512527466,
      "epoch": 6.23,
      "learning_rate": 1.5004043988696647e-06,
      "loss": 0.0541,
      "step": 6558,
      "task_loss": 0.13419125974178314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989131243884782,
      "compression_loss": 0.0,
      "distillation_loss": 0.0402006097137928,
      "epoch": 6.23,
      "learning_rate": 1.4967707610200083e-06,
      "loss": 0.0407,
      "step": 6559,
      "task_loss": 0.0453285276889801
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798914870840653,
      "compression_loss": 0.0,
      "distillation_loss": 0.043260738253593445,
      "epoch": 6.23,
      "learning_rate": 1.493141392645131e-06,
      "loss": 0.0406,
      "step": 6560,
      "task_loss": 0.01651701144874096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989166154209608,
      "compression_loss": 0.0,
      "distillation_loss": 0.1227022111415863,
      "epoch": 6.23,
      "learning_rate": 1.4895162944043334e-06,
      "loss": 0.1221,
      "step": 6561,
      "task_loss": 0.11656280606985092
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989183581304051,
      "compression_loss": 0.0,
      "distillation_loss": 0.01965763419866562,
      "epoch": 6.23,
      "learning_rate": 1.4858954669561275e-06,
      "loss": 0.0259,
      "step": 6562,
      "task_loss": 0.08250312507152557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989200989699897,
      "compression_loss": 0.0,
      "distillation_loss": 0.03354266285896301,
      "epoch": 6.23,
      "learning_rate": 1.4822789109582513e-06,
      "loss": 0.031,
      "step": 6563,
      "task_loss": 0.00812080129981041
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989218379407182,
      "compression_loss": 0.0,
      "distillation_loss": 0.10104814171791077,
      "epoch": 6.23,
      "learning_rate": 1.478666627067679e-06,
      "loss": 0.0967,
      "step": 6564,
      "task_loss": 0.05804811045527458
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989235750435945,
      "compression_loss": 0.0,
      "distillation_loss": 0.07657065242528915,
      "epoch": 6.23,
      "learning_rate": 1.4750586159405915e-06,
      "loss": 0.0822,
      "step": 6565,
      "task_loss": 0.13330647349357605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989253102796219,
      "compression_loss": 0.0,
      "distillation_loss": 0.018394935876131058,
      "epoch": 6.24,
      "learning_rate": 1.4714548782324034e-06,
      "loss": 0.0265,
      "step": 6566,
      "task_loss": 0.09906549006700516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989270436498043,
      "compression_loss": 0.0,
      "distillation_loss": 0.051937442272901535,
      "epoch": 6.24,
      "learning_rate": 1.467855414597749e-06,
      "loss": 0.0475,
      "step": 6567,
      "task_loss": 0.007465232163667679
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989287751551454,
      "compression_loss": 0.0,
      "distillation_loss": 0.049969661980867386,
      "epoch": 6.24,
      "learning_rate": 1.4642602256904946e-06,
      "loss": 0.0474,
      "step": 6568,
      "task_loss": 0.023928040638566017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798930504796649,
      "compression_loss": 0.0,
      "distillation_loss": 0.03371240198612213,
      "epoch": 6.24,
      "learning_rate": 1.4606693121637206e-06,
      "loss": 0.0316,
      "step": 6569,
      "task_loss": 0.012854812666773796
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989322325753184,
      "compression_loss": 0.0,
      "distillation_loss": 0.2045125961303711,
      "epoch": 6.24,
      "learning_rate": 1.457082674669727e-06,
      "loss": 0.2015,
      "step": 6570,
      "task_loss": 0.17422333359718323
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989339584921576,
      "compression_loss": 0.0,
      "distillation_loss": 0.02525508962571621,
      "epoch": 6.24,
      "learning_rate": 1.4535003138600566e-06,
      "loss": 0.0305,
      "step": 6571,
      "task_loss": 0.07814519852399826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989356825481703,
      "compression_loss": 0.0,
      "distillation_loss": 0.07085863500833511,
      "epoch": 6.24,
      "learning_rate": 1.4499222303854532e-06,
      "loss": 0.0728,
      "step": 6572,
      "task_loss": 0.09030604362487793
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989374047443599,
      "compression_loss": 0.0,
      "distillation_loss": 0.09062013030052185,
      "epoch": 6.24,
      "learning_rate": 1.4463484248958908e-06,
      "loss": 0.0856,
      "step": 6573,
      "task_loss": 0.04006872698664665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989391250817305,
      "compression_loss": 0.0,
      "distillation_loss": 0.0634981244802475,
      "epoch": 6.24,
      "learning_rate": 1.4427788980405728e-06,
      "loss": 0.0603,
      "step": 6574,
      "task_loss": 0.031479597091674805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989408435612854,
      "compression_loss": 0.0,
      "distillation_loss": 0.0360356867313385,
      "epoch": 6.24,
      "learning_rate": 1.4392136504679244e-06,
      "loss": 0.0351,
      "step": 6575,
      "task_loss": 0.026443321257829666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989425601840284,
      "compression_loss": 0.0,
      "distillation_loss": 0.0444665402173996,
      "epoch": 6.25,
      "learning_rate": 1.4356526828255862e-06,
      "loss": 0.0505,
      "step": 6576,
      "task_loss": 0.10513627529144287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989442749509632,
      "compression_loss": 0.0,
      "distillation_loss": 0.026376407593488693,
      "epoch": 6.25,
      "learning_rate": 1.432095995760424e-06,
      "loss": 0.034,
      "step": 6577,
      "task_loss": 0.10233157873153687
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989459878630936,
      "compression_loss": 0.0,
      "distillation_loss": 0.04820052161812782,
      "epoch": 6.25,
      "learning_rate": 1.4285435899185295e-06,
      "loss": 0.0516,
      "step": 6578,
      "task_loss": 0.08200520277023315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989476989214231,
      "compression_loss": 0.0,
      "distillation_loss": 0.06717506051063538,
      "epoch": 6.25,
      "learning_rate": 1.424995465945214e-06,
      "loss": 0.0792,
      "step": 6579,
      "task_loss": 0.1877632588148117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989494081269555,
      "compression_loss": 0.0,
      "distillation_loss": 0.05002850666642189,
      "epoch": 6.25,
      "learning_rate": 1.4214516244850068e-06,
      "loss": 0.0534,
      "step": 6580,
      "task_loss": 0.08330188691616058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989511154806944,
      "compression_loss": 0.0,
      "distillation_loss": 0.055970218032598495,
      "epoch": 6.25,
      "learning_rate": 1.417912066181673e-06,
      "loss": 0.0618,
      "step": 6581,
      "task_loss": 0.11391516029834747
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989528209836435,
      "compression_loss": 0.0,
      "distillation_loss": 0.03707553446292877,
      "epoch": 6.25,
      "learning_rate": 1.414376791678182e-06,
      "loss": 0.045,
      "step": 6582,
      "task_loss": 0.11639431864023209
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989545246368066,
      "compression_loss": 0.0,
      "distillation_loss": 0.09972971677780151,
      "epoch": 6.25,
      "learning_rate": 1.4108458016167337e-06,
      "loss": 0.1031,
      "step": 6583,
      "task_loss": 0.13307958841323853
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989562264411872,
      "compression_loss": 0.0,
      "distillation_loss": 0.15906114876270294,
      "epoch": 6.25,
      "learning_rate": 1.407319096638754e-06,
      "loss": 0.16,
      "step": 6584,
      "task_loss": 0.16835139691829681
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989579263977891,
      "compression_loss": 0.0,
      "distillation_loss": 0.24002781510353088,
      "epoch": 6.25,
      "learning_rate": 1.403796677384886e-06,
      "loss": 0.2357,
      "step": 6585,
      "task_loss": 0.19643017649650574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989596245076159,
      "compression_loss": 0.0,
      "distillation_loss": 0.023733289912343025,
      "epoch": 6.25,
      "learning_rate": 1.4002785444949928e-06,
      "loss": 0.0217,
      "step": 6586,
      "task_loss": 0.003847165033221245
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989613207716714,
      "compression_loss": 0.0,
      "distillation_loss": 0.028031958267092705,
      "epoch": 6.26,
      "learning_rate": 1.396764698608155e-06,
      "loss": 0.0294,
      "step": 6587,
      "task_loss": 0.04165419936180115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989630151909591,
      "compression_loss": 0.0,
      "distillation_loss": 0.0592382550239563,
      "epoch": 6.26,
      "learning_rate": 1.393255140362687e-06,
      "loss": 0.0561,
      "step": 6588,
      "task_loss": 0.027590807527303696
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798964707766483,
      "compression_loss": 0.0,
      "distillation_loss": 0.04887683689594269,
      "epoch": 6.26,
      "learning_rate": 1.3897498703961148e-06,
      "loss": 0.047,
      "step": 6589,
      "task_loss": 0.030398232862353325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989663984992464,
      "compression_loss": 0.0,
      "distillation_loss": 0.12447142601013184,
      "epoch": 6.26,
      "learning_rate": 1.3862488893451847e-06,
      "loss": 0.1271,
      "step": 6590,
      "task_loss": 0.15074971318244934
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989680873902533,
      "compression_loss": 0.0,
      "distillation_loss": 0.04007259011268616,
      "epoch": 6.26,
      "learning_rate": 1.3827521978458713e-06,
      "loss": 0.0504,
      "step": 6591,
      "task_loss": 0.14287948608398438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989697744405071,
      "compression_loss": 0.0,
      "distillation_loss": 0.05581611022353172,
      "epoch": 6.26,
      "learning_rate": 1.3792597965333581e-06,
      "loss": 0.0642,
      "step": 6592,
      "task_loss": 0.13964907824993134
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989714596510117,
      "compression_loss": 0.0,
      "distillation_loss": 0.024131378158926964,
      "epoch": 6.26,
      "learning_rate": 1.3757716860420683e-06,
      "loss": 0.0303,
      "step": 6593,
      "task_loss": 0.08553370833396912
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989731430227707,
      "compression_loss": 0.0,
      "distillation_loss": 0.020088283345103264,
      "epoch": 6.26,
      "learning_rate": 1.3722878670056227e-06,
      "loss": 0.0185,
      "step": 6594,
      "task_loss": 0.003951072692871094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989748245567877,
      "compression_loss": 0.0,
      "distillation_loss": 0.031761474907398224,
      "epoch": 6.26,
      "learning_rate": 1.368808340056879e-06,
      "loss": 0.0333,
      "step": 6595,
      "task_loss": 0.04679109528660774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989765042540665,
      "compression_loss": 0.0,
      "distillation_loss": 0.02415706217288971,
      "epoch": 6.26,
      "learning_rate": 1.3653331058279122e-06,
      "loss": 0.0225,
      "step": 6596,
      "task_loss": 0.007262144237756729
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989781821156109,
      "compression_loss": 0.0,
      "distillation_loss": 0.0280197374522686,
      "epoch": 6.26,
      "learning_rate": 1.361862164950009e-06,
      "loss": 0.0287,
      "step": 6597,
      "task_loss": 0.03486378863453865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989798581424242,
      "compression_loss": 0.0,
      "distillation_loss": 0.015901973471045494,
      "epoch": 6.27,
      "learning_rate": 1.35839551805369e-06,
      "loss": 0.0156,
      "step": 6598,
      "task_loss": 0.012705270200967789
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989815323355105,
      "compression_loss": 0.0,
      "distillation_loss": 0.024427423253655434,
      "epoch": 6.27,
      "learning_rate": 1.354933165768682e-06,
      "loss": 0.0236,
      "step": 6599,
      "task_loss": 0.016438543796539307
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989832046958733,
      "compression_loss": 0.0,
      "distillation_loss": 0.05082686245441437,
      "epoch": 6.27,
      "learning_rate": 1.3514751087239402e-06,
      "loss": 0.0588,
      "step": 6600,
      "task_loss": 0.13012255728244781
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989848752245162,
      "compression_loss": 0.0,
      "distillation_loss": 0.019267812371253967,
      "epoch": 6.27,
      "learning_rate": 1.3480213475476344e-06,
      "loss": 0.026,
      "step": 6601,
      "task_loss": 0.08636897802352905
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798986543922443,
      "compression_loss": 0.0,
      "distillation_loss": 0.09770554304122925,
      "epoch": 6.27,
      "learning_rate": 1.3445718828671655e-06,
      "loss": 0.0972,
      "step": 6602,
      "task_loss": 0.09245851635932922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989882107906573,
      "compression_loss": 0.0,
      "distillation_loss": 0.03061073273420334,
      "epoch": 6.27,
      "learning_rate": 1.3411267153091378e-06,
      "loss": 0.0339,
      "step": 6603,
      "task_loss": 0.06348737329244614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989898758301629,
      "compression_loss": 0.0,
      "distillation_loss": 0.09166794270277023,
      "epoch": 6.27,
      "learning_rate": 1.3376858454993813e-06,
      "loss": 0.1133,
      "step": 6604,
      "task_loss": 0.30779021978378296
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989915390419633,
      "compression_loss": 0.0,
      "distillation_loss": 0.030499190092086792,
      "epoch": 6.27,
      "learning_rate": 1.3342492740629541e-06,
      "loss": 0.0326,
      "step": 6605,
      "task_loss": 0.05136849731206894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989932004270623,
      "compression_loss": 0.0,
      "distillation_loss": 0.05024800822138786,
      "epoch": 6.27,
      "learning_rate": 1.330817001624124e-06,
      "loss": 0.0474,
      "step": 6606,
      "task_loss": 0.02216871827840805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989948599864637,
      "compression_loss": 0.0,
      "distillation_loss": 0.029151551425457,
      "epoch": 6.27,
      "learning_rate": 1.3273890288063722e-06,
      "loss": 0.0349,
      "step": 6607,
      "task_loss": 0.08693638443946838
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.798996517721171,
      "compression_loss": 0.0,
      "distillation_loss": 0.08503906428813934,
      "epoch": 6.28,
      "learning_rate": 1.323965356232415e-06,
      "loss": 0.0872,
      "step": 6608,
      "task_loss": 0.10668846219778061
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989981736321878,
      "compression_loss": 0.0,
      "distillation_loss": 0.03245946019887924,
      "epoch": 6.28,
      "learning_rate": 1.3205459845241714e-06,
      "loss": 0.0299,
      "step": 6609,
      "task_loss": 0.006885243579745293
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7989998277205181,
      "compression_loss": 0.0,
      "distillation_loss": 0.0519396997988224,
      "epoch": 6.28,
      "learning_rate": 1.317130914302797e-06,
      "loss": 0.051,
      "step": 6610,
      "task_loss": 0.04283105209469795
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990014799871654,
      "compression_loss": 0.0,
      "distillation_loss": 0.03950023651123047,
      "epoch": 6.28,
      "learning_rate": 1.3137201461886434e-06,
      "loss": 0.0457,
      "step": 6611,
      "task_loss": 0.10137784481048584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990031304331333,
      "compression_loss": 0.0,
      "distillation_loss": 0.03535441309213638,
      "epoch": 6.28,
      "learning_rate": 1.3103136808013061e-06,
      "loss": 0.0381,
      "step": 6612,
      "task_loss": 0.06290829181671143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990047790594256,
      "compression_loss": 0.0,
      "distillation_loss": 0.07202986627817154,
      "epoch": 6.28,
      "learning_rate": 1.3069115187595793e-06,
      "loss": 0.0809,
      "step": 6613,
      "task_loss": 0.1610376387834549
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799006425867046,
      "compression_loss": 0.0,
      "distillation_loss": 0.128701314330101,
      "epoch": 6.28,
      "learning_rate": 1.3035136606814769e-06,
      "loss": 0.1386,
      "step": 6614,
      "task_loss": 0.22806209325790405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799008070856998,
      "compression_loss": 0.0,
      "distillation_loss": 0.05727306380867958,
      "epoch": 6.28,
      "learning_rate": 1.3001201071842466e-06,
      "loss": 0.0639,
      "step": 6615,
      "task_loss": 0.1234469786286354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990097140302854,
      "compression_loss": 0.0,
      "distillation_loss": 0.029445184394717216,
      "epoch": 6.28,
      "learning_rate": 1.2967308588843375e-06,
      "loss": 0.027,
      "step": 6616,
      "task_loss": 0.0048278942704200745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799011355387912,
      "compression_loss": 0.0,
      "distillation_loss": 0.06337439268827438,
      "epoch": 6.28,
      "learning_rate": 1.2933459163974203e-06,
      "loss": 0.077,
      "step": 6617,
      "task_loss": 0.19938617944717407
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990129949308814,
      "compression_loss": 0.0,
      "distillation_loss": 0.03333733230829239,
      "epoch": 6.28,
      "learning_rate": 1.2899652803383926e-06,
      "loss": 0.0305,
      "step": 6618,
      "task_loss": 0.00532466359436512
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990146326601971,
      "compression_loss": 0.0,
      "distillation_loss": 0.07982062548398972,
      "epoch": 6.29,
      "learning_rate": 1.286588951321363e-06,
      "loss": 0.0878,
      "step": 6619,
      "task_loss": 0.1597793698310852
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990162685768631,
      "compression_loss": 0.0,
      "distillation_loss": 0.02913430519402027,
      "epoch": 6.29,
      "learning_rate": 1.2832169299596546e-06,
      "loss": 0.027,
      "step": 6620,
      "task_loss": 0.00752607174217701
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990179026818829,
      "compression_loss": 0.0,
      "distillation_loss": 0.05205439403653145,
      "epoch": 6.29,
      "learning_rate": 1.2798492168658083e-06,
      "loss": 0.0606,
      "step": 6621,
      "task_loss": 0.13726334273815155
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990195349762601,
      "compression_loss": 0.0,
      "distillation_loss": 0.050472185015678406,
      "epoch": 6.29,
      "learning_rate": 1.2764858126515928e-06,
      "loss": 0.0563,
      "step": 6622,
      "task_loss": 0.10921429097652435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990211654609986,
      "compression_loss": 0.0,
      "distillation_loss": 0.031548745930194855,
      "epoch": 6.29,
      "learning_rate": 1.2731267179279832e-06,
      "loss": 0.0319,
      "step": 6623,
      "task_loss": 0.03549543768167496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799022794137102,
      "compression_loss": 0.0,
      "distillation_loss": 0.06983719766139984,
      "epoch": 6.29,
      "learning_rate": 1.2697719333051723e-06,
      "loss": 0.082,
      "step": 6624,
      "task_loss": 0.191563218832016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990244210055738,
      "compression_loss": 0.0,
      "distillation_loss": 0.019796200096607208,
      "epoch": 6.29,
      "learning_rate": 1.2664214593925776e-06,
      "loss": 0.0262,
      "step": 6625,
      "task_loss": 0.08367738872766495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990260460674179,
      "compression_loss": 0.0,
      "distillation_loss": 0.038845110684633255,
      "epoch": 6.29,
      "learning_rate": 1.263075296798824e-06,
      "loss": 0.052,
      "step": 6626,
      "task_loss": 0.17067357897758484
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990276693236379,
      "compression_loss": 0.0,
      "distillation_loss": 0.029467133805155754,
      "epoch": 6.29,
      "learning_rate": 1.2597334461317667e-06,
      "loss": 0.0382,
      "step": 6627,
      "task_loss": 0.11635088920593262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990292907752375,
      "compression_loss": 0.0,
      "distillation_loss": 0.0395752415060997,
      "epoch": 6.29,
      "learning_rate": 1.2563959079984588e-06,
      "loss": 0.037,
      "step": 6628,
      "task_loss": 0.014232108369469643
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990309104232205,
      "compression_loss": 0.0,
      "distillation_loss": 0.024423949420452118,
      "epoch": 6.3,
      "learning_rate": 1.2530626830051878e-06,
      "loss": 0.0302,
      "step": 6629,
      "task_loss": 0.08202096819877625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990325282685903,
      "compression_loss": 0.0,
      "distillation_loss": 0.03499576449394226,
      "epoch": 6.3,
      "learning_rate": 1.249733771757447e-06,
      "loss": 0.0361,
      "step": 6630,
      "task_loss": 0.045992329716682434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990341443123508,
      "compression_loss": 0.0,
      "distillation_loss": 0.11989525705575943,
      "epoch": 6.3,
      "learning_rate": 1.2464091748599443e-06,
      "loss": 0.1271,
      "step": 6631,
      "task_loss": 0.19175073504447937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990357585555057,
      "compression_loss": 0.0,
      "distillation_loss": 0.030376242473721504,
      "epoch": 6.3,
      "learning_rate": 1.243088892916619e-06,
      "loss": 0.0319,
      "step": 6632,
      "task_loss": 0.04513759911060333
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990373709990585,
      "compression_loss": 0.0,
      "distillation_loss": 0.10441594570875168,
      "epoch": 6.3,
      "learning_rate": 1.239772926530608e-06,
      "loss": 0.1108,
      "step": 6633,
      "task_loss": 0.16776102781295776
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799038981644013,
      "compression_loss": 0.0,
      "distillation_loss": 0.04732033982872963,
      "epoch": 6.3,
      "learning_rate": 1.2364612763042793e-06,
      "loss": 0.0544,
      "step": 6634,
      "task_loss": 0.11861524730920792
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990405904913729,
      "compression_loss": 0.0,
      "distillation_loss": 0.022771336138248444,
      "epoch": 6.3,
      "learning_rate": 1.2331539428391963e-06,
      "loss": 0.021,
      "step": 6635,
      "task_loss": 0.005372023209929466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990421975421418,
      "compression_loss": 0.0,
      "distillation_loss": 0.06504938006401062,
      "epoch": 6.3,
      "learning_rate": 1.2298509267361702e-06,
      "loss": 0.0743,
      "step": 6636,
      "task_loss": 0.1580170840024948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990438027973236,
      "compression_loss": 0.0,
      "distillation_loss": 0.04712830111384392,
      "epoch": 6.3,
      "learning_rate": 1.2265522285952013e-06,
      "loss": 0.051,
      "step": 6637,
      "task_loss": 0.08559815585613251
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990454062579216,
      "compression_loss": 0.0,
      "distillation_loss": 0.024217963218688965,
      "epoch": 6.3,
      "learning_rate": 1.2232578490155105e-06,
      "loss": 0.0226,
      "step": 6638,
      "task_loss": 0.0075660087168216705
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990470079249398,
      "compression_loss": 0.0,
      "distillation_loss": 0.021058350801467896,
      "epoch": 6.3,
      "learning_rate": 1.219967788595544e-06,
      "loss": 0.0246,
      "step": 6639,
      "task_loss": 0.05632723867893219
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990486077993818,
      "compression_loss": 0.0,
      "distillation_loss": 0.056007929146289825,
      "epoch": 6.31,
      "learning_rate": 1.2166820479329572e-06,
      "loss": 0.0745,
      "step": 6640,
      "task_loss": 0.24108853936195374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990502058822513,
      "compression_loss": 0.0,
      "distillation_loss": 0.02392977848649025,
      "epoch": 6.31,
      "learning_rate": 1.2134006276246169e-06,
      "loss": 0.0225,
      "step": 6641,
      "task_loss": 0.00946834497153759
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990518021745519,
      "compression_loss": 0.0,
      "distillation_loss": 0.027623049914836884,
      "epoch": 6.31,
      "learning_rate": 1.2101235282666045e-06,
      "loss": 0.026,
      "step": 6642,
      "task_loss": 0.011274173855781555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990533966772873,
      "compression_loss": 0.0,
      "distillation_loss": 0.04122261703014374,
      "epoch": 6.31,
      "learning_rate": 1.2068507504542332e-06,
      "loss": 0.0483,
      "step": 6643,
      "task_loss": 0.11227674782276154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990549893914611,
      "compression_loss": 0.0,
      "distillation_loss": 0.014780269004404545,
      "epoch": 6.31,
      "learning_rate": 1.2035822947820074e-06,
      "loss": 0.0215,
      "step": 6644,
      "task_loss": 0.08238159120082855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990565803180774,
      "compression_loss": 0.0,
      "distillation_loss": 0.10595270246267319,
      "epoch": 6.31,
      "learning_rate": 1.200318161843661e-06,
      "loss": 0.1129,
      "step": 6645,
      "task_loss": 0.17518508434295654
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990581694581393,
      "compression_loss": 0.0,
      "distillation_loss": 0.04436454176902771,
      "epoch": 6.31,
      "learning_rate": 1.1970583522321472e-06,
      "loss": 0.0475,
      "step": 6646,
      "task_loss": 0.07612089067697525
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990597568126508,
      "compression_loss": 0.0,
      "distillation_loss": 0.028508765622973442,
      "epoch": 6.31,
      "learning_rate": 1.1938028665396173e-06,
      "loss": 0.0261,
      "step": 6647,
      "task_loss": 0.003986643627285957
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990613423826156,
      "compression_loss": 0.0,
      "distillation_loss": 0.12394885718822479,
      "epoch": 6.31,
      "learning_rate": 1.190551705357451e-06,
      "loss": 0.1199,
      "step": 6648,
      "task_loss": 0.08359791338443756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990629261690374,
      "compression_loss": 0.0,
      "distillation_loss": 0.0263461172580719,
      "epoch": 6.31,
      "learning_rate": 1.187304869276229e-06,
      "loss": 0.0244,
      "step": 6649,
      "task_loss": 0.006703455001115799
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990645081729196,
      "compression_loss": 0.0,
      "distillation_loss": 0.04055074229836464,
      "epoch": 6.32,
      "learning_rate": 1.184062358885768e-06,
      "loss": 0.0377,
      "step": 6650,
      "task_loss": 0.012343864887952805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990660883952663,
      "compression_loss": 0.0,
      "distillation_loss": 0.03743916004896164,
      "epoch": 6.32,
      "learning_rate": 1.1808241747750748e-06,
      "loss": 0.0488,
      "step": 6651,
      "task_loss": 0.1512567400932312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990676668370809,
      "compression_loss": 0.0,
      "distillation_loss": 0.05502176284790039,
      "epoch": 6.32,
      "learning_rate": 1.1775903175323787e-06,
      "loss": 0.0604,
      "step": 6652,
      "task_loss": 0.10876131802797318
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799069243499367,
      "compression_loss": 0.0,
      "distillation_loss": 0.026486551389098167,
      "epoch": 6.32,
      "learning_rate": 1.174360787745138e-06,
      "loss": 0.0318,
      "step": 6653,
      "task_loss": 0.07932649552822113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990708183831287,
      "compression_loss": 0.0,
      "distillation_loss": 0.09362354874610901,
      "epoch": 6.32,
      "learning_rate": 1.1711355860000079e-06,
      "loss": 0.091,
      "step": 6654,
      "task_loss": 0.06723588705062866
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990723914893691,
      "compression_loss": 0.0,
      "distillation_loss": 0.13072776794433594,
      "epoch": 6.32,
      "learning_rate": 1.167914712882856e-06,
      "loss": 0.134,
      "step": 6655,
      "task_loss": 0.1634681522846222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990739628190924,
      "compression_loss": 0.0,
      "distillation_loss": 0.014383634552359581,
      "epoch": 6.32,
      "learning_rate": 1.1646981689787728e-06,
      "loss": 0.0228,
      "step": 6656,
      "task_loss": 0.09892392158508301
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990755323733021,
      "compression_loss": 0.0,
      "distillation_loss": 0.013216648250818253,
      "epoch": 6.32,
      "learning_rate": 1.1614859548720603e-06,
      "loss": 0.0251,
      "step": 6657,
      "task_loss": 0.13216069340705872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990771001530017,
      "compression_loss": 0.0,
      "distillation_loss": 0.06353971362113953,
      "epoch": 6.32,
      "learning_rate": 1.1582780711462321e-06,
      "loss": 0.0629,
      "step": 6658,
      "task_loss": 0.057146959006786346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990786661591952,
      "compression_loss": 0.0,
      "distillation_loss": 0.06731215119361877,
      "epoch": 6.32,
      "learning_rate": 1.1550745183840139e-06,
      "loss": 0.0659,
      "step": 6659,
      "task_loss": 0.05347077175974846
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799080230392886,
      "compression_loss": 0.0,
      "distillation_loss": 0.018060829490423203,
      "epoch": 6.32,
      "learning_rate": 1.1518752971673485e-06,
      "loss": 0.0169,
      "step": 6660,
      "task_loss": 0.006073372438549995
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799081792855078,
      "compression_loss": 0.0,
      "distillation_loss": 0.05580337345600128,
      "epoch": 6.33,
      "learning_rate": 1.1486804080773877e-06,
      "loss": 0.0684,
      "step": 6661,
      "task_loss": 0.18163037300109863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990833535467747,
      "compression_loss": 0.0,
      "distillation_loss": 0.07261084020137787,
      "epoch": 6.33,
      "learning_rate": 1.1454898516945035e-06,
      "loss": 0.0715,
      "step": 6662,
      "task_loss": 0.061981115490198135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79908491246898,
      "compression_loss": 0.0,
      "distillation_loss": 0.031433992087841034,
      "epoch": 6.33,
      "learning_rate": 1.142303628598268e-06,
      "loss": 0.0484,
      "step": 6663,
      "task_loss": 0.20125338435173035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990864696226974,
      "compression_loss": 0.0,
      "distillation_loss": 0.057327330112457275,
      "epoch": 6.33,
      "learning_rate": 1.1391217393674825e-06,
      "loss": 0.0524,
      "step": 6664,
      "task_loss": 0.008000411093235016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990880250089306,
      "compression_loss": 0.0,
      "distillation_loss": 0.02260660007596016,
      "epoch": 6.33,
      "learning_rate": 1.1359441845801483e-06,
      "loss": 0.0213,
      "step": 6665,
      "task_loss": 0.00976894237101078
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990895786286834,
      "compression_loss": 0.0,
      "distillation_loss": 0.019381048157811165,
      "epoch": 6.33,
      "learning_rate": 1.1327709648134787e-06,
      "loss": 0.0255,
      "step": 6666,
      "task_loss": 0.08072338998317719
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990911304829592,
      "compression_loss": 0.0,
      "distillation_loss": 0.042968470603227615,
      "epoch": 6.33,
      "learning_rate": 1.1296020806439128e-06,
      "loss": 0.0399,
      "step": 6667,
      "task_loss": 0.012241264805197716
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990926805727621,
      "compression_loss": 0.0,
      "distillation_loss": 0.05627443641424179,
      "epoch": 6.33,
      "learning_rate": 1.1264375326470926e-06,
      "loss": 0.0537,
      "step": 6668,
      "task_loss": 0.030838757753372192
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990942288990954,
      "compression_loss": 0.0,
      "distillation_loss": 0.043274518102407455,
      "epoch": 6.33,
      "learning_rate": 1.1232773213978642e-06,
      "loss": 0.0483,
      "step": 6669,
      "task_loss": 0.09389711171388626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799095775462963,
      "compression_loss": 0.0,
      "distillation_loss": 0.027110418304800987,
      "epoch": 6.33,
      "learning_rate": 1.1201214474703043e-06,
      "loss": 0.0259,
      "step": 6670,
      "task_loss": 0.014662139117717743
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990973202653685,
      "compression_loss": 0.0,
      "distillation_loss": 0.022032059729099274,
      "epoch": 6.34,
      "learning_rate": 1.1169699114376931e-06,
      "loss": 0.031,
      "step": 6671,
      "task_loss": 0.11220581084489822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7990988633073157,
      "compression_loss": 0.0,
      "distillation_loss": 0.014706656336784363,
      "epoch": 6.34,
      "learning_rate": 1.1138227138725171e-06,
      "loss": 0.0135,
      "step": 6672,
      "task_loss": 0.0031169112771749496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991004045898081,
      "compression_loss": 0.0,
      "distillation_loss": 0.030809402465820312,
      "epoch": 6.34,
      "learning_rate": 1.1106798553464804e-06,
      "loss": 0.0434,
      "step": 6673,
      "task_loss": 0.15683190524578094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991019441138495,
      "compression_loss": 0.0,
      "distillation_loss": 0.12929940223693848,
      "epoch": 6.34,
      "learning_rate": 1.1075413364305037e-06,
      "loss": 0.1301,
      "step": 6674,
      "task_loss": 0.1377396285533905
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991034818804436,
      "compression_loss": 0.0,
      "distillation_loss": 0.017488310113549232,
      "epoch": 6.34,
      "learning_rate": 1.1044071576947118e-06,
      "loss": 0.016,
      "step": 6675,
      "task_loss": 0.0023362338542938232
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799105017890594,
      "compression_loss": 0.0,
      "distillation_loss": 0.024973999708890915,
      "epoch": 6.34,
      "learning_rate": 1.101277319708438e-06,
      "loss": 0.0233,
      "step": 6676,
      "task_loss": 0.007980858907103539
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991065521453045,
      "compression_loss": 0.0,
      "distillation_loss": 0.0375574491918087,
      "epoch": 6.34,
      "learning_rate": 1.0981518230402387e-06,
      "loss": 0.0393,
      "step": 6677,
      "task_loss": 0.05460818111896515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991080846455786,
      "compression_loss": 0.0,
      "distillation_loss": 0.04282548278570175,
      "epoch": 6.34,
      "learning_rate": 1.0950306682578709e-06,
      "loss": 0.041,
      "step": 6678,
      "task_loss": 0.024096237495541573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991096153924201,
      "compression_loss": 0.0,
      "distillation_loss": 0.08127694576978683,
      "epoch": 6.34,
      "learning_rate": 1.0919138559283143e-06,
      "loss": 0.0851,
      "step": 6679,
      "task_loss": 0.11989080905914307
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991111443868326,
      "compression_loss": 0.0,
      "distillation_loss": 0.025165293365716934,
      "epoch": 6.34,
      "learning_rate": 1.0888013866177437e-06,
      "loss": 0.037,
      "step": 6680,
      "task_loss": 0.1433994174003601
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991126716298199,
      "compression_loss": 0.0,
      "distillation_loss": 0.027159132063388824,
      "epoch": 6.34,
      "learning_rate": 1.0856932608915627e-06,
      "loss": 0.0298,
      "step": 6681,
      "task_loss": 0.05401374772191048
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991141971223857,
      "compression_loss": 0.0,
      "distillation_loss": 0.13656893372535706,
      "epoch": 6.35,
      "learning_rate": 1.0825894793143721e-06,
      "loss": 0.1396,
      "step": 6682,
      "task_loss": 0.16644863784313202
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991157208655336,
      "compression_loss": 0.0,
      "distillation_loss": 0.05064243823289871,
      "epoch": 6.35,
      "learning_rate": 1.0794900424499876e-06,
      "loss": 0.0506,
      "step": 6683,
      "task_loss": 0.05057719722390175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991172428602672,
      "compression_loss": 0.0,
      "distillation_loss": 0.07734186947345734,
      "epoch": 6.35,
      "learning_rate": 1.0763949508614423e-06,
      "loss": 0.0739,
      "step": 6684,
      "task_loss": 0.04286087676882744
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991187631075903,
      "compression_loss": 0.0,
      "distillation_loss": 0.08190207928419113,
      "epoch": 6.35,
      "learning_rate": 1.0733042051109726e-06,
      "loss": 0.0776,
      "step": 6685,
      "task_loss": 0.038602665066719055
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991202816085067,
      "compression_loss": 0.0,
      "distillation_loss": 0.04311536252498627,
      "epoch": 6.35,
      "learning_rate": 1.070217805760021e-06,
      "loss": 0.0555,
      "step": 6686,
      "task_loss": 0.16687297821044922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991217983640198,
      "compression_loss": 0.0,
      "distillation_loss": 0.03107578307390213,
      "epoch": 6.35,
      "learning_rate": 1.0671357533692554e-06,
      "loss": 0.0346,
      "step": 6687,
      "task_loss": 0.06593281030654907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991233133751334,
      "compression_loss": 0.0,
      "distillation_loss": 0.023783832788467407,
      "epoch": 6.35,
      "learning_rate": 1.0640580484985424e-06,
      "loss": 0.027,
      "step": 6688,
      "task_loss": 0.05620795860886574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991248266428512,
      "compression_loss": 0.0,
      "distillation_loss": 0.09371946007013321,
      "epoch": 6.35,
      "learning_rate": 1.0609846917069622e-06,
      "loss": 0.088,
      "step": 6689,
      "task_loss": 0.03661830723285675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799126338168177,
      "compression_loss": 0.0,
      "distillation_loss": 0.22859099507331848,
      "epoch": 6.35,
      "learning_rate": 1.0579156835528015e-06,
      "loss": 0.2157,
      "step": 6690,
      "task_loss": 0.09983595460653305
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991278479521143,
      "compression_loss": 0.0,
      "distillation_loss": 0.0681435763835907,
      "epoch": 6.35,
      "learning_rate": 1.0548510245935673e-06,
      "loss": 0.0716,
      "step": 6691,
      "task_loss": 0.10267331451177597
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991293559956669,
      "compression_loss": 0.0,
      "distillation_loss": 0.029233068227767944,
      "epoch": 6.36,
      "learning_rate": 1.051790715385964e-06,
      "loss": 0.0305,
      "step": 6692,
      "task_loss": 0.041752371937036514
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991308622998384,
      "compression_loss": 0.0,
      "distillation_loss": 0.02068510465323925,
      "epoch": 6.36,
      "learning_rate": 1.0487347564859113e-06,
      "loss": 0.0214,
      "step": 6693,
      "task_loss": 0.02766001597046852
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991323668656325,
      "compression_loss": 0.0,
      "distillation_loss": 0.009869755245745182,
      "epoch": 6.36,
      "learning_rate": 1.0456831484485423e-06,
      "loss": 0.0093,
      "step": 6694,
      "task_loss": 0.003924252465367317
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991338696940529,
      "compression_loss": 0.0,
      "distillation_loss": 0.06949397921562195,
      "epoch": 6.36,
      "learning_rate": 1.0426358918281948e-06,
      "loss": 0.0649,
      "step": 6695,
      "task_loss": 0.023511435836553574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991353707861033,
      "compression_loss": 0.0,
      "distillation_loss": 0.03212964907288551,
      "epoch": 6.36,
      "learning_rate": 1.0395929871784144e-06,
      "loss": 0.0365,
      "step": 6696,
      "task_loss": 0.07626573741436005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991368701427873,
      "compression_loss": 0.0,
      "distillation_loss": 0.08683188259601593,
      "epoch": 6.36,
      "learning_rate": 1.0365544350519646e-06,
      "loss": 0.0829,
      "step": 6697,
      "task_loss": 0.047511257231235504
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991383677651087,
      "compression_loss": 0.0,
      "distillation_loss": 0.031793013215065,
      "epoch": 6.36,
      "learning_rate": 1.0335202360008124e-06,
      "loss": 0.0338,
      "step": 6698,
      "task_loss": 0.051480717957019806
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991398636540712,
      "compression_loss": 0.0,
      "distillation_loss": 0.024651937186717987,
      "epoch": 6.36,
      "learning_rate": 1.0304903905761332e-06,
      "loss": 0.0227,
      "step": 6699,
      "task_loss": 0.0050739627331495285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991413578106783,
      "compression_loss": 0.0,
      "distillation_loss": 0.03288833424448967,
      "epoch": 6.36,
      "learning_rate": 1.0274648993283093e-06,
      "loss": 0.0333,
      "step": 6700,
      "task_loss": 0.03704278543591499
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991428502359338,
      "compression_loss": 0.0,
      "distillation_loss": 0.07430990785360336,
      "epoch": 6.36,
      "learning_rate": 1.0244437628069425e-06,
      "loss": 0.0758,
      "step": 6701,
      "task_loss": 0.08946188539266586
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991443409308414,
      "compression_loss": 0.0,
      "distillation_loss": 0.031692154705524445,
      "epoch": 6.36,
      "learning_rate": 1.0214269815608358e-06,
      "loss": 0.0424,
      "step": 6702,
      "task_loss": 0.1384189873933792
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991458298964047,
      "compression_loss": 0.0,
      "distillation_loss": 0.03650377690792084,
      "epoch": 6.37,
      "learning_rate": 1.018414556137995e-06,
      "loss": 0.0546,
      "step": 6703,
      "task_loss": 0.21728603541851044
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991473171336275,
      "compression_loss": 0.0,
      "distillation_loss": 0.14218676090240479,
      "epoch": 6.37,
      "learning_rate": 1.015406487085646e-06,
      "loss": 0.1351,
      "step": 6704,
      "task_loss": 0.07109387218952179
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991488026435134,
      "compression_loss": 0.0,
      "distillation_loss": 0.08769318461418152,
      "epoch": 6.37,
      "learning_rate": 1.0124027749502246e-06,
      "loss": 0.092,
      "step": 6705,
      "task_loss": 0.13085408508777618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991502864270661,
      "compression_loss": 0.0,
      "distillation_loss": 0.0496244877576828,
      "epoch": 6.37,
      "learning_rate": 1.0094034202773634e-06,
      "loss": 0.0473,
      "step": 6706,
      "task_loss": 0.026335952803492546
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991517684852892,
      "compression_loss": 0.0,
      "distillation_loss": 0.058904923498630524,
      "epoch": 6.37,
      "learning_rate": 1.006408423611907e-06,
      "loss": 0.0619,
      "step": 6707,
      "task_loss": 0.08892402052879333
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991532488191866,
      "compression_loss": 0.0,
      "distillation_loss": 0.03160509839653969,
      "epoch": 6.37,
      "learning_rate": 1.0034177854979205e-06,
      "loss": 0.0358,
      "step": 6708,
      "task_loss": 0.07310265302658081
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991547274297618,
      "compression_loss": 0.0,
      "distillation_loss": 0.02190888673067093,
      "epoch": 6.37,
      "learning_rate": 1.0004315064786608e-06,
      "loss": 0.0332,
      "step": 6709,
      "task_loss": 0.134353369474411
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991562043180185,
      "compression_loss": 0.0,
      "distillation_loss": 0.02776992879807949,
      "epoch": 6.37,
      "learning_rate": 9.974495870965967e-07,
      "loss": 0.0338,
      "step": 6710,
      "task_loss": 0.08767490833997726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991576794849604,
      "compression_loss": 0.0,
      "distillation_loss": 0.03906933218240738,
      "epoch": 6.37,
      "learning_rate": 9.944720278934171e-07,
      "loss": 0.0357,
      "step": 6711,
      "task_loss": 0.005863867700099945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991591529315912,
      "compression_loss": 0.0,
      "distillation_loss": 0.05178176239132881,
      "epoch": 6.37,
      "learning_rate": 9.914988294100063e-07,
      "loss": 0.0492,
      "step": 6712,
      "task_loss": 0.026018494740128517
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991606246589146,
      "compression_loss": 0.0,
      "distillation_loss": 0.018913403153419495,
      "epoch": 6.38,
      "learning_rate": 9.885299921864543e-07,
      "loss": 0.0217,
      "step": 6713,
      "task_loss": 0.04700911417603493
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991620946679343,
      "compression_loss": 0.0,
      "distillation_loss": 0.045953452587127686,
      "epoch": 6.38,
      "learning_rate": 9.855655167620715e-07,
      "loss": 0.0522,
      "step": 6714,
      "task_loss": 0.10820292681455612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991635629596539,
      "compression_loss": 0.0,
      "distillation_loss": 0.09742830693721771,
      "epoch": 6.38,
      "learning_rate": 9.826054036753713e-07,
      "loss": 0.092,
      "step": 6715,
      "task_loss": 0.04326290637254715
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991650295350771,
      "compression_loss": 0.0,
      "distillation_loss": 0.03743236884474754,
      "epoch": 6.38,
      "learning_rate": 9.796496534640653e-07,
      "loss": 0.0345,
      "step": 6716,
      "task_loss": 0.008360574021935463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991664943952076,
      "compression_loss": 0.0,
      "distillation_loss": 0.060953132808208466,
      "epoch": 6.38,
      "learning_rate": 9.766982666650826e-07,
      "loss": 0.0634,
      "step": 6717,
      "task_loss": 0.08522552996873856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991679575410491,
      "compression_loss": 0.0,
      "distillation_loss": 0.047717705368995667,
      "epoch": 6.38,
      "learning_rate": 9.737512438145579e-07,
      "loss": 0.053,
      "step": 6718,
      "task_loss": 0.10008193552494049
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991694189736052,
      "compression_loss": 0.0,
      "distillation_loss": 0.024083293974399567,
      "epoch": 6.38,
      "learning_rate": 9.708085854478327e-07,
      "loss": 0.022,
      "step": 6719,
      "task_loss": 0.0036613382399082184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991708786938797,
      "compression_loss": 0.0,
      "distillation_loss": 0.05113540589809418,
      "epoch": 6.38,
      "learning_rate": 9.678702920994543e-07,
      "loss": 0.0567,
      "step": 6720,
      "task_loss": 0.1063266173005104
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991723367028762,
      "compression_loss": 0.0,
      "distillation_loss": 0.028987891972064972,
      "epoch": 6.38,
      "learning_rate": 9.649363643031733e-07,
      "loss": 0.0268,
      "step": 6721,
      "task_loss": 0.007400134578347206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991737930015985,
      "compression_loss": 0.0,
      "distillation_loss": 0.04050924628973007,
      "epoch": 6.38,
      "learning_rate": 9.620068025919583e-07,
      "loss": 0.0462,
      "step": 6722,
      "task_loss": 0.097807377576828
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991752475910502,
      "compression_loss": 0.0,
      "distillation_loss": 0.0265050008893013,
      "epoch": 6.38,
      "learning_rate": 9.590816074979774e-07,
      "loss": 0.0258,
      "step": 6723,
      "task_loss": 0.019627349451184273
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991767004722349,
      "compression_loss": 0.0,
      "distillation_loss": 0.10371723771095276,
      "epoch": 6.39,
      "learning_rate": 9.561607795526007e-07,
      "loss": 0.1172,
      "step": 6724,
      "task_loss": 0.2384086400270462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991781516461564,
      "compression_loss": 0.0,
      "distillation_loss": 0.033512018620967865,
      "epoch": 6.39,
      "learning_rate": 9.532443192864199e-07,
      "loss": 0.0316,
      "step": 6725,
      "task_loss": 0.013961471617221832
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991796011138184,
      "compression_loss": 0.0,
      "distillation_loss": 0.03146098554134369,
      "epoch": 6.39,
      "learning_rate": 9.50332227229217e-07,
      "loss": 0.0535,
      "step": 6726,
      "task_loss": 0.2520584762096405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991810488762244,
      "compression_loss": 0.0,
      "distillation_loss": 0.026955943554639816,
      "epoch": 6.39,
      "learning_rate": 9.474245039099882e-07,
      "loss": 0.0253,
      "step": 6727,
      "task_loss": 0.010826632380485535
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991824949343782,
      "compression_loss": 0.0,
      "distillation_loss": 0.024500368162989616,
      "epoch": 6.39,
      "learning_rate": 9.445211498569362e-07,
      "loss": 0.0334,
      "step": 6728,
      "task_loss": 0.11393891274929047
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991839392892836,
      "compression_loss": 0.0,
      "distillation_loss": 0.05089277774095535,
      "epoch": 6.39,
      "learning_rate": 9.416221655974722e-07,
      "loss": 0.0473,
      "step": 6729,
      "task_loss": 0.014700580388307571
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991853819419441,
      "compression_loss": 0.0,
      "distillation_loss": 0.01643327623605728,
      "epoch": 6.39,
      "learning_rate": 9.387275516582056e-07,
      "loss": 0.0304,
      "step": 6730,
      "task_loss": 0.1562596559524536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991868228933635,
      "compression_loss": 0.0,
      "distillation_loss": 0.07891779392957687,
      "epoch": 6.39,
      "learning_rate": 9.358373085649602e-07,
      "loss": 0.0943,
      "step": 6731,
      "task_loss": 0.23281759023666382
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991882621445454,
      "compression_loss": 0.0,
      "distillation_loss": 0.020287442952394485,
      "epoch": 6.39,
      "learning_rate": 9.329514368427633e-07,
      "loss": 0.0234,
      "step": 6732,
      "task_loss": 0.051706451922655106
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991896996964936,
      "compression_loss": 0.0,
      "distillation_loss": 0.0532609187066555,
      "epoch": 6.39,
      "learning_rate": 9.300699370158456e-07,
      "loss": 0.0599,
      "step": 6733,
      "task_loss": 0.11954209208488464
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991911355502115,
      "compression_loss": 0.0,
      "distillation_loss": 0.037625864148139954,
      "epoch": 6.4,
      "learning_rate": 9.271928096076493e-07,
      "loss": 0.0411,
      "step": 6734,
      "task_loss": 0.07242215424776077
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991925697067032,
      "compression_loss": 0.0,
      "distillation_loss": 0.02129533141851425,
      "epoch": 6.4,
      "learning_rate": 9.243200551408094e-07,
      "loss": 0.0389,
      "step": 6735,
      "task_loss": 0.19778040051460266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799194002166972,
      "compression_loss": 0.0,
      "distillation_loss": 0.030612638220191002,
      "epoch": 6.4,
      "learning_rate": 9.214516741371831e-07,
      "loss": 0.0464,
      "step": 6736,
      "task_loss": 0.18832482397556305
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991954329320218,
      "compression_loss": 0.0,
      "distillation_loss": 0.01869923248887062,
      "epoch": 6.4,
      "learning_rate": 9.185876671178262e-07,
      "loss": 0.0174,
      "step": 6737,
      "task_loss": 0.005990633741021156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991968620028562,
      "compression_loss": 0.0,
      "distillation_loss": 0.02486245334148407,
      "epoch": 6.4,
      "learning_rate": 9.157280346029918e-07,
      "loss": 0.0232,
      "step": 6738,
      "task_loss": 0.00861707329750061
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799198289380479,
      "compression_loss": 0.0,
      "distillation_loss": 0.03501253202557564,
      "epoch": 6.4,
      "learning_rate": 9.128727771121531e-07,
      "loss": 0.0323,
      "step": 6739,
      "task_loss": 0.007670162245631218
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7991997150658937,
      "compression_loss": 0.0,
      "distillation_loss": 0.041302017867565155,
      "epoch": 6.4,
      "learning_rate": 9.100218951639816e-07,
      "loss": 0.0531,
      "step": 6740,
      "task_loss": 0.15916121006011963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992011390601041,
      "compression_loss": 0.0,
      "distillation_loss": 0.056890472769737244,
      "epoch": 6.4,
      "learning_rate": 9.071753892763519e-07,
      "loss": 0.0589,
      "step": 6741,
      "task_loss": 0.0772441178560257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992025613641139,
      "compression_loss": 0.0,
      "distillation_loss": 0.02093365229666233,
      "epoch": 6.4,
      "learning_rate": 9.043332599663418e-07,
      "loss": 0.0277,
      "step": 6742,
      "task_loss": 0.08839461952447891
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992039819789267,
      "compression_loss": 0.0,
      "distillation_loss": 0.09711463004350662,
      "epoch": 6.4,
      "learning_rate": 9.014955077502413e-07,
      "loss": 0.1142,
      "step": 6743,
      "task_loss": 0.268157958984375
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992054009055463,
      "compression_loss": 0.0,
      "distillation_loss": 0.0171291995793581,
      "epoch": 6.4,
      "learning_rate": 8.986621331435435e-07,
      "loss": 0.0159,
      "step": 6744,
      "task_loss": 0.004604209214448929
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992068181449762,
      "compression_loss": 0.0,
      "distillation_loss": 0.09282074868679047,
      "epoch": 6.41,
      "learning_rate": 8.958331366609423e-07,
      "loss": 0.0897,
      "step": 6745,
      "task_loss": 0.061169348657131195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992082336982201,
      "compression_loss": 0.0,
      "distillation_loss": 0.07672721892595291,
      "epoch": 6.41,
      "learning_rate": 8.930085188163378e-07,
      "loss": 0.08,
      "step": 6746,
      "task_loss": 0.1096218079328537
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799209647566282,
      "compression_loss": 0.0,
      "distillation_loss": 0.02508268877863884,
      "epoch": 6.41,
      "learning_rate": 8.90188280122839e-07,
      "loss": 0.0238,
      "step": 6747,
      "task_loss": 0.012722663581371307
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992110597501652,
      "compression_loss": 0.0,
      "distillation_loss": 0.024517200887203217,
      "epoch": 6.41,
      "learning_rate": 8.87372421092747e-07,
      "loss": 0.0225,
      "step": 6748,
      "task_loss": 0.00447007454931736
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992124702508735,
      "compression_loss": 0.0,
      "distillation_loss": 0.01891644299030304,
      "epoch": 6.41,
      "learning_rate": 8.845609422375861e-07,
      "loss": 0.0175,
      "step": 6749,
      "task_loss": 0.00444909930229187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992138790694107,
      "compression_loss": 0.0,
      "distillation_loss": 0.02350194752216339,
      "epoch": 6.41,
      "learning_rate": 8.817538440680728e-07,
      "loss": 0.0224,
      "step": 6750,
      "task_loss": 0.01221693865954876
    },
    {
      "epoch": 6.41,
      "eval_accuracy": 0.8910550458715596,
      "eval_loss": 0.43947499990463257,
      "eval_runtime": 18.2778,
      "eval_samples_per_second": 47.708,
      "eval_steps_per_second": 5.964,
      "step": 6750
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992152862067803,
      "compression_loss": 0.0,
      "distillation_loss": 0.03163065388798714,
      "epoch": 6.41,
      "learning_rate": 8.78951127094127e-07,
      "loss": 0.0336,
      "step": 6751,
      "task_loss": 0.05086345970630646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992166916639861,
      "compression_loss": 0.0,
      "distillation_loss": 0.10874065011739731,
      "epoch": 6.41,
      "learning_rate": 8.761527918248775e-07,
      "loss": 0.1034,
      "step": 6752,
      "task_loss": 0.05520922690629959
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992180954420318,
      "compression_loss": 0.0,
      "distillation_loss": 0.02465911954641342,
      "epoch": 6.41,
      "learning_rate": 8.733588387686537e-07,
      "loss": 0.0227,
      "step": 6753,
      "task_loss": 0.005142947658896446
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992194975419211,
      "compression_loss": 0.0,
      "distillation_loss": 0.07918893545866013,
      "epoch": 6.41,
      "learning_rate": 8.705692684329969e-07,
      "loss": 0.0822,
      "step": 6754,
      "task_loss": 0.10922683775424957
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992208979646576,
      "compression_loss": 0.0,
      "distillation_loss": 0.030382784083485603,
      "epoch": 6.42,
      "learning_rate": 8.677840813246352e-07,
      "loss": 0.0294,
      "step": 6755,
      "task_loss": 0.020388955250382423
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992222967112449,
      "compression_loss": 0.0,
      "distillation_loss": 0.035736992955207825,
      "epoch": 6.42,
      "learning_rate": 8.650032779495165e-07,
      "loss": 0.0331,
      "step": 6756,
      "task_loss": 0.009340623393654823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992236937826869,
      "compression_loss": 0.0,
      "distillation_loss": 0.045183077454566956,
      "epoch": 6.42,
      "learning_rate": 8.622268588127924e-07,
      "loss": 0.0426,
      "step": 6757,
      "task_loss": 0.019841017201542854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799225089179987,
      "compression_loss": 0.0,
      "distillation_loss": 0.0189533494412899,
      "epoch": 6.42,
      "learning_rate": 8.594548244188067e-07,
      "loss": 0.0177,
      "step": 6758,
      "task_loss": 0.006474364548921585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992264829041492,
      "compression_loss": 0.0,
      "distillation_loss": 0.024439578875899315,
      "epoch": 6.42,
      "learning_rate": 8.56687175271112e-07,
      "loss": 0.0229,
      "step": 6759,
      "task_loss": 0.008742136880755424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992278749561771,
      "compression_loss": 0.0,
      "distillation_loss": 0.026206394657492638,
      "epoch": 6.42,
      "learning_rate": 8.539239118724701e-07,
      "loss": 0.0266,
      "step": 6760,
      "task_loss": 0.03009379468858242
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992292653370742,
      "compression_loss": 0.0,
      "distillation_loss": 0.016620401293039322,
      "epoch": 6.42,
      "learning_rate": 8.511650347248406e-07,
      "loss": 0.017,
      "step": 6761,
      "task_loss": 0.02082774229347706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992306540478442,
      "compression_loss": 0.0,
      "distillation_loss": 0.025677714496850967,
      "epoch": 6.42,
      "learning_rate": 8.484105443293783e-07,
      "loss": 0.0237,
      "step": 6762,
      "task_loss": 0.005807174369692802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992320410894911,
      "compression_loss": 0.0,
      "distillation_loss": 0.024803977459669113,
      "epoch": 6.42,
      "learning_rate": 8.456604411864605e-07,
      "loss": 0.0229,
      "step": 6763,
      "task_loss": 0.005409408360719681
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992334264630182,
      "compression_loss": 0.0,
      "distillation_loss": 0.09612073004245758,
      "epoch": 6.42,
      "learning_rate": 8.429147257956516e-07,
      "loss": 0.0983,
      "step": 6764,
      "task_loss": 0.11760664731264114
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992348101694294,
      "compression_loss": 0.0,
      "distillation_loss": 0.05043857544660568,
      "epoch": 6.42,
      "learning_rate": 8.401733986557247e-07,
      "loss": 0.0493,
      "step": 6765,
      "task_loss": 0.03950589895248413
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992361922097283,
      "compression_loss": 0.0,
      "distillation_loss": 0.05412200465798378,
      "epoch": 6.43,
      "learning_rate": 8.374364602646511e-07,
      "loss": 0.0535,
      "step": 6766,
      "task_loss": 0.04812926799058914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992375725849187,
      "compression_loss": 0.0,
      "distillation_loss": 0.02848942205309868,
      "epoch": 6.43,
      "learning_rate": 8.347039111196164e-07,
      "loss": 0.0328,
      "step": 6767,
      "task_loss": 0.07136546820402145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992389512960041,
      "compression_loss": 0.0,
      "distillation_loss": 0.11072133481502533,
      "epoch": 6.43,
      "learning_rate": 8.319757517169985e-07,
      "loss": 0.12,
      "step": 6768,
      "task_loss": 0.2039954513311386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992403283439883,
      "compression_loss": 0.0,
      "distillation_loss": 0.02107694186270237,
      "epoch": 6.43,
      "learning_rate": 8.29251982552376e-07,
      "loss": 0.0194,
      "step": 6769,
      "task_loss": 0.004783786833286285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992417037298749,
      "compression_loss": 0.0,
      "distillation_loss": 0.013939835131168365,
      "epoch": 6.43,
      "learning_rate": 8.265326041205417e-07,
      "loss": 0.0129,
      "step": 6770,
      "task_loss": 0.0032312143594026566
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992430774546677,
      "compression_loss": 0.0,
      "distillation_loss": 0.04151005297899246,
      "epoch": 6.43,
      "learning_rate": 8.238176169154816e-07,
      "loss": 0.0465,
      "step": 6771,
      "task_loss": 0.09118230640888214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992444495193702,
      "compression_loss": 0.0,
      "distillation_loss": 0.06526434421539307,
      "epoch": 6.43,
      "learning_rate": 8.211070214303812e-07,
      "loss": 0.0745,
      "step": 6772,
      "task_loss": 0.15805087983608246
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992458199249863,
      "compression_loss": 0.0,
      "distillation_loss": 0.024169165641069412,
      "epoch": 6.43,
      "learning_rate": 8.184008181576386e-07,
      "loss": 0.0227,
      "step": 6773,
      "task_loss": 0.009067356586456299
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992471886725196,
      "compression_loss": 0.0,
      "distillation_loss": 0.029119737446308136,
      "epoch": 6.43,
      "learning_rate": 8.156990075888466e-07,
      "loss": 0.0281,
      "step": 6774,
      "task_loss": 0.0189868975430727
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992485557629738,
      "compression_loss": 0.0,
      "distillation_loss": 0.03186260536313057,
      "epoch": 6.43,
      "learning_rate": 8.130015902148042e-07,
      "loss": 0.0347,
      "step": 6775,
      "task_loss": 0.06021621823310852
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992499211973525,
      "compression_loss": 0.0,
      "distillation_loss": 0.02992023155093193,
      "epoch": 6.43,
      "learning_rate": 8.103085665255084e-07,
      "loss": 0.0543,
      "step": 6776,
      "task_loss": 0.27328985929489136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992512849766594,
      "compression_loss": 0.0,
      "distillation_loss": 0.04072340205311775,
      "epoch": 6.44,
      "learning_rate": 8.076199370101594e-07,
      "loss": 0.0409,
      "step": 6777,
      "task_loss": 0.042335450649261475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992526471018983,
      "compression_loss": 0.0,
      "distillation_loss": 0.04745135456323624,
      "epoch": 6.44,
      "learning_rate": 8.04935702157164e-07,
      "loss": 0.0502,
      "step": 6778,
      "task_loss": 0.07529226690530777
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992540075740727,
      "compression_loss": 0.0,
      "distillation_loss": 0.04864822328090668,
      "epoch": 6.44,
      "learning_rate": 8.022558624541182e-07,
      "loss": 0.0524,
      "step": 6779,
      "task_loss": 0.08609342575073242
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992553663941864,
      "compression_loss": 0.0,
      "distillation_loss": 0.11904314160346985,
      "epoch": 6.44,
      "learning_rate": 7.995804183878353e-07,
      "loss": 0.1204,
      "step": 6780,
      "task_loss": 0.13219860196113586
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992567235632431,
      "compression_loss": 0.0,
      "distillation_loss": 0.0507618710398674,
      "epoch": 6.44,
      "learning_rate": 7.969093704443209e-07,
      "loss": 0.0556,
      "step": 6781,
      "task_loss": 0.09945084899663925
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992580790822463,
      "compression_loss": 0.0,
      "distillation_loss": 0.08536731451749802,
      "epoch": 6.44,
      "learning_rate": 7.942427191087786e-07,
      "loss": 0.0833,
      "step": 6782,
      "task_loss": 0.06510312855243683
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992594329521999,
      "compression_loss": 0.0,
      "distillation_loss": 0.09828724712133408,
      "epoch": 6.44,
      "learning_rate": 7.915804648656239e-07,
      "loss": 0.1016,
      "step": 6783,
      "task_loss": 0.13165351748466492
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992607851741076,
      "compression_loss": 0.0,
      "distillation_loss": 0.06299099326133728,
      "epoch": 6.44,
      "learning_rate": 7.889226081984696e-07,
      "loss": 0.0704,
      "step": 6784,
      "task_loss": 0.13664023578166962
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992621357489729,
      "compression_loss": 0.0,
      "distillation_loss": 0.017018694430589676,
      "epoch": 6.44,
      "learning_rate": 7.862691495901243e-07,
      "loss": 0.0163,
      "step": 6785,
      "task_loss": 0.00940592773258686
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992634846777995,
      "compression_loss": 0.0,
      "distillation_loss": 0.08001522719860077,
      "epoch": 6.44,
      "learning_rate": 7.83620089522602e-07,
      "loss": 0.0758,
      "step": 6786,
      "task_loss": 0.038091909140348434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992648319615913,
      "compression_loss": 0.0,
      "distillation_loss": 0.013534367084503174,
      "epoch": 6.45,
      "learning_rate": 7.809754284771181e-07,
      "loss": 0.0144,
      "step": 6787,
      "task_loss": 0.0219335425645113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992661776013518,
      "compression_loss": 0.0,
      "distillation_loss": 0.03589925542473793,
      "epoch": 6.45,
      "learning_rate": 7.783351669340882e-07,
      "loss": 0.0332,
      "step": 6788,
      "task_loss": 0.009313993155956268
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992675215980846,
      "compression_loss": 0.0,
      "distillation_loss": 0.044496770948171616,
      "epoch": 6.45,
      "learning_rate": 7.756993053731259e-07,
      "loss": 0.0418,
      "step": 6789,
      "task_loss": 0.01771704852581024
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992688639527935,
      "compression_loss": 0.0,
      "distillation_loss": 0.025459101423621178,
      "epoch": 6.45,
      "learning_rate": 7.730678442730538e-07,
      "loss": 0.0239,
      "step": 6790,
      "task_loss": 0.009527930989861488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992702046664822,
      "compression_loss": 0.0,
      "distillation_loss": 0.02458195760846138,
      "epoch": 6.45,
      "learning_rate": 7.704407841118811e-07,
      "loss": 0.0229,
      "step": 6791,
      "task_loss": 0.007813390344381332
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992715437401545,
      "compression_loss": 0.0,
      "distillation_loss": 0.01990782469511032,
      "epoch": 6.45,
      "learning_rate": 7.678181253668343e-07,
      "loss": 0.0228,
      "step": 6792,
      "task_loss": 0.04860430955886841
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992728811748139,
      "compression_loss": 0.0,
      "distillation_loss": 0.024153951555490494,
      "epoch": 6.45,
      "learning_rate": 7.651998685143269e-07,
      "loss": 0.0354,
      "step": 6793,
      "task_loss": 0.13703292608261108
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799274216971464,
      "compression_loss": 0.0,
      "distillation_loss": 0.02501821331679821,
      "epoch": 6.45,
      "learning_rate": 7.625860140299811e-07,
      "loss": 0.0231,
      "step": 6794,
      "task_loss": 0.006082385778427124
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992755511311087,
      "compression_loss": 0.0,
      "distillation_loss": 0.06825748831033707,
      "epoch": 6.45,
      "learning_rate": 7.599765623886146e-07,
      "loss": 0.0673,
      "step": 6795,
      "task_loss": 0.05868315324187279
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992768836547516,
      "compression_loss": 0.0,
      "distillation_loss": 0.040932249277830124,
      "epoch": 6.45,
      "learning_rate": 7.573715140642451e-07,
      "loss": 0.0558,
      "step": 6796,
      "task_loss": 0.18955527245998383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992782145433963,
      "compression_loss": 0.0,
      "distillation_loss": 0.03785280883312225,
      "epoch": 6.45,
      "learning_rate": 7.547708695300942e-07,
      "loss": 0.0502,
      "step": 6797,
      "task_loss": 0.16129517555236816
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992795437980466,
      "compression_loss": 0.0,
      "distillation_loss": 0.02329355478286743,
      "epoch": 6.46,
      "learning_rate": 7.521746292585841e-07,
      "loss": 0.0357,
      "step": 6798,
      "task_loss": 0.1471756100654602
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799280871419706,
      "compression_loss": 0.0,
      "distillation_loss": 0.03262361139059067,
      "epoch": 6.46,
      "learning_rate": 7.49582793721329e-07,
      "loss": 0.0336,
      "step": 6799,
      "task_loss": 0.04221324622631073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992821974093786,
      "compression_loss": 0.0,
      "distillation_loss": 0.03289227932691574,
      "epoch": 6.46,
      "learning_rate": 7.469953633891469e-07,
      "loss": 0.0401,
      "step": 6800,
      "task_loss": 0.10484490543603897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992835217680675,
      "compression_loss": 0.0,
      "distillation_loss": 0.07555963844060898,
      "epoch": 6.46,
      "learning_rate": 7.444123387320645e-07,
      "loss": 0.0751,
      "step": 6801,
      "task_loss": 0.07083379477262497
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992848444967768,
      "compression_loss": 0.0,
      "distillation_loss": 0.03143364191055298,
      "epoch": 6.46,
      "learning_rate": 7.418337202192982e-07,
      "loss": 0.0313,
      "step": 6802,
      "task_loss": 0.03013058938086033
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992861655965101,
      "compression_loss": 0.0,
      "distillation_loss": 0.03767804428935051,
      "epoch": 6.46,
      "learning_rate": 7.392595083192622e-07,
      "loss": 0.0348,
      "step": 6803,
      "task_loss": 0.009084422141313553
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992874850682711,
      "compression_loss": 0.0,
      "distillation_loss": 0.03183813393115997,
      "epoch": 6.46,
      "learning_rate": 7.366897034995796e-07,
      "loss": 0.0295,
      "step": 6804,
      "task_loss": 0.008527562022209167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992888029130634,
      "compression_loss": 0.0,
      "distillation_loss": 0.022205591201782227,
      "epoch": 6.46,
      "learning_rate": 7.341243062270686e-07,
      "loss": 0.028,
      "step": 6805,
      "task_loss": 0.0802656039595604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992901191318905,
      "compression_loss": 0.0,
      "distillation_loss": 0.036016229540109634,
      "epoch": 6.46,
      "learning_rate": 7.315633169677399e-07,
      "loss": 0.0352,
      "step": 6806,
      "task_loss": 0.027843181043863297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992914337257565,
      "compression_loss": 0.0,
      "distillation_loss": 0.03833125904202461,
      "epoch": 6.46,
      "learning_rate": 7.290067361868103e-07,
      "loss": 0.0388,
      "step": 6807,
      "task_loss": 0.0432337149977684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992927466956649,
      "compression_loss": 0.0,
      "distillation_loss": 0.01719411462545395,
      "epoch": 6.47,
      "learning_rate": 7.264545643486997e-07,
      "loss": 0.0272,
      "step": 6808,
      "task_loss": 0.11720463633537292
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992940580426192,
      "compression_loss": 0.0,
      "distillation_loss": 0.10508248209953308,
      "epoch": 6.47,
      "learning_rate": 7.239068019170209e-07,
      "loss": 0.1087,
      "step": 6809,
      "task_loss": 0.14166167378425598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992953677676233,
      "compression_loss": 0.0,
      "distillation_loss": 0.04110556095838547,
      "epoch": 6.47,
      "learning_rate": 7.21363449354584e-07,
      "loss": 0.0467,
      "step": 6810,
      "task_loss": 0.09688466787338257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992966758716809,
      "compression_loss": 0.0,
      "distillation_loss": 0.0683632344007492,
      "epoch": 6.47,
      "learning_rate": 7.188245071234057e-07,
      "loss": 0.0735,
      "step": 6811,
      "task_loss": 0.11992549151182175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7992979823557955,
      "compression_loss": 0.0,
      "distillation_loss": 0.03459140658378601,
      "epoch": 6.47,
      "learning_rate": 7.162899756846975e-07,
      "loss": 0.0356,
      "step": 6812,
      "task_loss": 0.04442333057522774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799299287220971,
      "compression_loss": 0.0,
      "distillation_loss": 0.05156092718243599,
      "epoch": 6.47,
      "learning_rate": 7.137598554988633e-07,
      "loss": 0.0484,
      "step": 6813,
      "task_loss": 0.019665861502289772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993005904682109,
      "compression_loss": 0.0,
      "distillation_loss": 0.07404929399490356,
      "epoch": 6.47,
      "learning_rate": 7.112341470255163e-07,
      "loss": 0.0795,
      "step": 6814,
      "task_loss": 0.12879236042499542
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993018920985189,
      "compression_loss": 0.0,
      "distillation_loss": 0.055263321846723557,
      "epoch": 6.47,
      "learning_rate": 7.087128507234642e-07,
      "loss": 0.0582,
      "step": 6815,
      "task_loss": 0.08427795022726059
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993031921128988,
      "compression_loss": 0.0,
      "distillation_loss": 0.06862097978591919,
      "epoch": 6.47,
      "learning_rate": 7.061959670507102e-07,
      "loss": 0.0758,
      "step": 6816,
      "task_loss": 0.1407826542854309
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993044905123541,
      "compression_loss": 0.0,
      "distillation_loss": 0.030200235545635223,
      "epoch": 6.47,
      "learning_rate": 7.036834964644523e-07,
      "loss": 0.0385,
      "step": 6817,
      "task_loss": 0.11273118853569031
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993057872978887,
      "compression_loss": 0.0,
      "distillation_loss": 0.015482475981116295,
      "epoch": 6.47,
      "learning_rate": 7.011754394211061e-07,
      "loss": 0.0144,
      "step": 6818,
      "task_loss": 0.004941888153553009
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993070824705062,
      "compression_loss": 0.0,
      "distillation_loss": 0.05421324819326401,
      "epoch": 6.48,
      "learning_rate": 6.986717963762656e-07,
      "loss": 0.0614,
      "step": 6819,
      "task_loss": 0.12607041001319885
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993083760312102,
      "compression_loss": 0.0,
      "distillation_loss": 0.05300523713231087,
      "epoch": 6.48,
      "learning_rate": 6.961725677847308e-07,
      "loss": 0.0534,
      "step": 6820,
      "task_loss": 0.0573628805577755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993096679810044,
      "compression_loss": 0.0,
      "distillation_loss": 0.031698837876319885,
      "epoch": 6.48,
      "learning_rate": 6.936777541004941e-07,
      "loss": 0.0293,
      "step": 6821,
      "task_loss": 0.007633762434124947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993109583208926,
      "compression_loss": 0.0,
      "distillation_loss": 0.053836889564991,
      "epoch": 6.48,
      "learning_rate": 6.911873557767568e-07,
      "loss": 0.0578,
      "step": 6822,
      "task_loss": 0.09317664802074432
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993122470518784,
      "compression_loss": 0.0,
      "distillation_loss": 0.02741752751171589,
      "epoch": 6.48,
      "learning_rate": 6.88701373265907e-07,
      "loss": 0.0269,
      "step": 6823,
      "task_loss": 0.022505706176161766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993135341749654,
      "compression_loss": 0.0,
      "distillation_loss": 0.05598681792616844,
      "epoch": 6.48,
      "learning_rate": 6.862198070195363e-07,
      "loss": 0.0687,
      "step": 6824,
      "task_loss": 0.18340638279914856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993148196911575,
      "compression_loss": 0.0,
      "distillation_loss": 0.060910556465387344,
      "epoch": 6.48,
      "learning_rate": 6.837426574884342e-07,
      "loss": 0.0622,
      "step": 6825,
      "task_loss": 0.07380431145429611
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993161036014581,
      "compression_loss": 0.0,
      "distillation_loss": 0.03215963393449783,
      "epoch": 6.48,
      "learning_rate": 6.812699251225907e-07,
      "loss": 0.0296,
      "step": 6826,
      "task_loss": 0.006164673715829849
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993173859068712,
      "compression_loss": 0.0,
      "distillation_loss": 0.0726788341999054,
      "epoch": 6.48,
      "learning_rate": 6.788016103711825e-07,
      "loss": 0.0694,
      "step": 6827,
      "task_loss": 0.039418820291757584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993186666084002,
      "compression_loss": 0.0,
      "distillation_loss": 0.042292624711990356,
      "epoch": 6.48,
      "learning_rate": 6.763377136825927e-07,
      "loss": 0.039,
      "step": 6828,
      "task_loss": 0.009175049141049385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993199457070489,
      "compression_loss": 0.0,
      "distillation_loss": 0.018559550866484642,
      "epoch": 6.49,
      "learning_rate": 6.738782355044049e-07,
      "loss": 0.0244,
      "step": 6829,
      "task_loss": 0.07653959095478058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799321223203821,
      "compression_loss": 0.0,
      "distillation_loss": 0.14147916436195374,
      "epoch": 6.49,
      "learning_rate": 6.714231762833894e-07,
      "loss": 0.1379,
      "step": 6830,
      "task_loss": 0.10579898953437805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993224990997202,
      "compression_loss": 0.0,
      "distillation_loss": 0.02243266999721527,
      "epoch": 6.49,
      "learning_rate": 6.689725364655203e-07,
      "loss": 0.0242,
      "step": 6831,
      "task_loss": 0.04005550593137741
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993237733957501,
      "compression_loss": 0.0,
      "distillation_loss": 0.0320717990398407,
      "epoch": 6.49,
      "learning_rate": 6.665263164959745e-07,
      "loss": 0.049,
      "step": 6832,
      "task_loss": 0.20118948817253113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993250460929144,
      "compression_loss": 0.0,
      "distillation_loss": 0.02864461950957775,
      "epoch": 6.49,
      "learning_rate": 6.640845168191107e-07,
      "loss": 0.0266,
      "step": 6833,
      "task_loss": 0.008286131545901299
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993263171922169,
      "compression_loss": 0.0,
      "distillation_loss": 0.014984526671469212,
      "epoch": 6.49,
      "learning_rate": 6.616471378784961e-07,
      "loss": 0.014,
      "step": 6834,
      "task_loss": 0.0047838687896728516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993275866946612,
      "compression_loss": 0.0,
      "distillation_loss": 0.0654832050204277,
      "epoch": 6.49,
      "learning_rate": 6.592141801168933e-07,
      "loss": 0.0741,
      "step": 6835,
      "task_loss": 0.15198421478271484
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993288546012509,
      "compression_loss": 0.0,
      "distillation_loss": 0.030285434797406197,
      "epoch": 6.49,
      "learning_rate": 6.567856439762654e-07,
      "loss": 0.0406,
      "step": 6836,
      "task_loss": 0.13304871320724487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993301209129897,
      "compression_loss": 0.0,
      "distillation_loss": 0.052962590008974075,
      "epoch": 6.49,
      "learning_rate": 6.543615298977623e-07,
      "loss": 0.0483,
      "step": 6837,
      "task_loss": 0.006451290100812912
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993313856308815,
      "compression_loss": 0.0,
      "distillation_loss": 0.07559768110513687,
      "epoch": 6.49,
      "learning_rate": 6.519418383217347e-07,
      "loss": 0.0742,
      "step": 6838,
      "task_loss": 0.061944808810949326
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993326487559297,
      "compression_loss": 0.0,
      "distillation_loss": 0.042924731969833374,
      "epoch": 6.49,
      "learning_rate": 6.495265696877361e-07,
      "loss": 0.0429,
      "step": 6839,
      "task_loss": 0.04244181513786316
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993339102891381,
      "compression_loss": 0.0,
      "distillation_loss": 0.021773064509034157,
      "epoch": 6.5,
      "learning_rate": 6.471157244345105e-07,
      "loss": 0.0202,
      "step": 6840,
      "task_loss": 0.005988283082842827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993351702315104,
      "compression_loss": 0.0,
      "distillation_loss": 0.01895114779472351,
      "epoch": 6.5,
      "learning_rate": 6.447093029999935e-07,
      "loss": 0.0179,
      "step": 6841,
      "task_loss": 0.008197702467441559
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993364285840503,
      "compression_loss": 0.0,
      "distillation_loss": 0.06650319695472717,
      "epoch": 6.5,
      "learning_rate": 6.423073058213325e-07,
      "loss": 0.0657,
      "step": 6842,
      "task_loss": 0.058126937597990036
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993376853477614,
      "compression_loss": 0.0,
      "distillation_loss": 0.030550938099622726,
      "epoch": 6.5,
      "learning_rate": 6.399097333348536e-07,
      "loss": 0.0346,
      "step": 6843,
      "task_loss": 0.0713091567158699
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993389405236474,
      "compression_loss": 0.0,
      "distillation_loss": 0.01258410420268774,
      "epoch": 6.5,
      "learning_rate": 6.375165859760946e-07,
      "loss": 0.0116,
      "step": 6844,
      "task_loss": 0.003225104883313179
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993401941127121,
      "compression_loss": 0.0,
      "distillation_loss": 0.04864715039730072,
      "epoch": 6.5,
      "learning_rate": 6.351278641797742e-07,
      "loss": 0.0461,
      "step": 6845,
      "task_loss": 0.023460451513528824
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799341446115959,
      "compression_loss": 0.0,
      "distillation_loss": 0.10957182943820953,
      "epoch": 6.5,
      "learning_rate": 6.327435683798233e-07,
      "loss": 0.1103,
      "step": 6846,
      "task_loss": 0.1172788143157959
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993426965343919,
      "compression_loss": 0.0,
      "distillation_loss": 0.050183918327093124,
      "epoch": 6.5,
      "learning_rate": 6.303636990093592e-07,
      "loss": 0.059,
      "step": 6847,
      "task_loss": 0.13854621350765228
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993439453690145,
      "compression_loss": 0.0,
      "distillation_loss": 0.013963685370981693,
      "epoch": 6.5,
      "learning_rate": 6.279882565006889e-07,
      "loss": 0.0201,
      "step": 6848,
      "task_loss": 0.07548151910305023
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993451926208305,
      "compression_loss": 0.0,
      "distillation_loss": 0.0308124627918005,
      "epoch": 6.5,
      "learning_rate": 6.256172412853339e-07,
      "loss": 0.0313,
      "step": 6849,
      "task_loss": 0.0358404815196991
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993464382908434,
      "compression_loss": 0.0,
      "distillation_loss": 0.006880389526486397,
      "epoch": 6.51,
      "learning_rate": 6.232506537939941e-07,
      "loss": 0.0065,
      "step": 6850,
      "task_loss": 0.00280972383916378
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799347682380057,
      "compression_loss": 0.0,
      "distillation_loss": 0.03846500813961029,
      "epoch": 6.51,
      "learning_rate": 6.208884944565702e-07,
      "loss": 0.0352,
      "step": 6851,
      "task_loss": 0.005746711045503616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993489248894751,
      "compression_loss": 0.0,
      "distillation_loss": 0.0295537281781435,
      "epoch": 6.51,
      "learning_rate": 6.185307637021631e-07,
      "loss": 0.0389,
      "step": 6852,
      "task_loss": 0.12259507924318314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993501658201012,
      "compression_loss": 0.0,
      "distillation_loss": 0.08173021674156189,
      "epoch": 6.51,
      "learning_rate": 6.161774619590666e-07,
      "loss": 0.0784,
      "step": 6853,
      "task_loss": 0.048187751322984695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799351405172939,
      "compression_loss": 0.0,
      "distillation_loss": 0.10911855846643448,
      "epoch": 6.51,
      "learning_rate": 6.138285896547691e-07,
      "loss": 0.1035,
      "step": 6854,
      "task_loss": 0.05309395492076874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993526429489923,
      "compression_loss": 0.0,
      "distillation_loss": 0.06665603071451187,
      "epoch": 6.51,
      "learning_rate": 6.114841472159516e-07,
      "loss": 0.0703,
      "step": 6855,
      "task_loss": 0.10340512543916702
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993538791492647,
      "compression_loss": 0.0,
      "distillation_loss": 0.013052813708782196,
      "epoch": 6.51,
      "learning_rate": 6.091441350684957e-07,
      "loss": 0.0121,
      "step": 6856,
      "task_loss": 0.003964599221944809
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993551137747599,
      "compression_loss": 0.0,
      "distillation_loss": 0.030018752440810204,
      "epoch": 6.51,
      "learning_rate": 6.068085536374752e-07,
      "loss": 0.0292,
      "step": 6857,
      "task_loss": 0.02144923247396946
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993563468264816,
      "compression_loss": 0.0,
      "distillation_loss": 0.020197410136461258,
      "epoch": 6.51,
      "learning_rate": 6.04477403347159e-07,
      "loss": 0.0321,
      "step": 6858,
      "task_loss": 0.13970831036567688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993575783054334,
      "compression_loss": 0.0,
      "distillation_loss": 0.052353113889694214,
      "epoch": 6.51,
      "learning_rate": 6.02150684621014e-07,
      "loss": 0.0531,
      "step": 6859,
      "task_loss": 0.059832729399204254
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993588082126191,
      "compression_loss": 0.0,
      "distillation_loss": 0.04445452615618706,
      "epoch": 6.51,
      "learning_rate": 5.998283978816966e-07,
      "loss": 0.0456,
      "step": 6860,
      "task_loss": 0.05596046522259712
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993600365490422,
      "compression_loss": 0.0,
      "distillation_loss": 0.10008314251899719,
      "epoch": 6.52,
      "learning_rate": 5.975105435510637e-07,
      "loss": 0.1065,
      "step": 6861,
      "task_loss": 0.1641978919506073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993612633157067,
      "compression_loss": 0.0,
      "distillation_loss": 0.02374127134680748,
      "epoch": 6.52,
      "learning_rate": 5.951971220501645e-07,
      "loss": 0.0222,
      "step": 6862,
      "task_loss": 0.008045762777328491
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993624885136159,
      "compression_loss": 0.0,
      "distillation_loss": 0.04164247214794159,
      "epoch": 6.52,
      "learning_rate": 5.928881337992437e-07,
      "loss": 0.0465,
      "step": 6863,
      "task_loss": 0.08977600187063217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993637121437738,
      "compression_loss": 0.0,
      "distillation_loss": 0.06360233575105667,
      "epoch": 6.52,
      "learning_rate": 5.905835792177406e-07,
      "loss": 0.088,
      "step": 6864,
      "task_loss": 0.30765628814697266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993649342071839,
      "compression_loss": 0.0,
      "distillation_loss": 0.03090558759868145,
      "epoch": 6.52,
      "learning_rate": 5.882834587242842e-07,
      "loss": 0.043,
      "step": 6865,
      "task_loss": 0.15227064490318298
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79936615470485,
      "compression_loss": 0.0,
      "distillation_loss": 0.02005811221897602,
      "epoch": 6.52,
      "learning_rate": 5.859877727367069e-07,
      "loss": 0.0293,
      "step": 6866,
      "task_loss": 0.11207105964422226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993673736377757,
      "compression_loss": 0.0,
      "distillation_loss": 0.06416010856628418,
      "epoch": 6.52,
      "learning_rate": 5.836965216720309e-07,
      "loss": 0.0746,
      "step": 6867,
      "task_loss": 0.16828285157680511
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993685910069646,
      "compression_loss": 0.0,
      "distillation_loss": 0.09754236787557602,
      "epoch": 6.52,
      "learning_rate": 5.814097059464702e-07,
      "loss": 0.0941,
      "step": 6868,
      "task_loss": 0.06287850439548492
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993698068134205,
      "compression_loss": 0.0,
      "distillation_loss": 0.016108129173517227,
      "epoch": 6.52,
      "learning_rate": 5.79127325975437e-07,
      "loss": 0.0156,
      "step": 6869,
      "task_loss": 0.010948818176984787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993710210581472,
      "compression_loss": 0.0,
      "distillation_loss": 0.04941096901893616,
      "epoch": 6.52,
      "learning_rate": 5.768493821735387e-07,
      "loss": 0.0568,
      "step": 6870,
      "task_loss": 0.12317119538784027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799372233742148,
      "compression_loss": 0.0,
      "distillation_loss": 0.06292635947465897,
      "epoch": 6.53,
      "learning_rate": 5.745758749545749e-07,
      "loss": 0.0747,
      "step": 6871,
      "task_loss": 0.18059583008289337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799373444866427,
      "compression_loss": 0.0,
      "distillation_loss": 0.028696933761239052,
      "epoch": 6.53,
      "learning_rate": 5.723068047315344e-07,
      "loss": 0.0358,
      "step": 6872,
      "task_loss": 0.09973844140768051
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993746544319877,
      "compression_loss": 0.0,
      "distillation_loss": 0.10228434950113297,
      "epoch": 6.53,
      "learning_rate": 5.7004217191661e-07,
      "loss": 0.0967,
      "step": 6873,
      "task_loss": 0.04644036293029785
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993758624398338,
      "compression_loss": 0.0,
      "distillation_loss": 0.05492546781897545,
      "epoch": 6.53,
      "learning_rate": 5.677819769211807e-07,
      "loss": 0.0512,
      "step": 6874,
      "task_loss": 0.01718452386558056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799377068890969,
      "compression_loss": 0.0,
      "distillation_loss": 0.0156172476708889,
      "epoch": 6.53,
      "learning_rate": 5.655262201558209e-07,
      "loss": 0.0151,
      "step": 6875,
      "task_loss": 0.010073995217680931
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993782737863969,
      "compression_loss": 0.0,
      "distillation_loss": 0.04025673121213913,
      "epoch": 6.53,
      "learning_rate": 5.63274902030303e-07,
      "loss": 0.0425,
      "step": 6876,
      "task_loss": 0.06272687762975693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993794771271213,
      "compression_loss": 0.0,
      "distillation_loss": 0.11300121247768402,
      "epoch": 6.53,
      "learning_rate": 5.610280229535858e-07,
      "loss": 0.1092,
      "step": 6877,
      "task_loss": 0.07530970871448517
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993806789141458,
      "compression_loss": 0.0,
      "distillation_loss": 0.03029126301407814,
      "epoch": 6.53,
      "learning_rate": 5.58785583333829e-07,
      "loss": 0.0353,
      "step": 6878,
      "task_loss": 0.08056721091270447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799381879148474,
      "compression_loss": 0.0,
      "distillation_loss": 0.02235785312950611,
      "epoch": 6.53,
      "learning_rate": 5.56547583578379e-07,
      "loss": 0.0225,
      "step": 6879,
      "task_loss": 0.023482179269194603
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993830778311097,
      "compression_loss": 0.0,
      "distillation_loss": 0.026940112933516502,
      "epoch": 6.53,
      "learning_rate": 5.543140240937827e-07,
      "loss": 0.0271,
      "step": 6880,
      "task_loss": 0.02874702401459217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993842749630566,
      "compression_loss": 0.0,
      "distillation_loss": 0.03313375264406204,
      "epoch": 6.53,
      "learning_rate": 5.520849052857768e-07,
      "loss": 0.0304,
      "step": 6881,
      "task_loss": 0.005804285407066345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993854705453185,
      "compression_loss": 0.0,
      "distillation_loss": 0.05707018822431564,
      "epoch": 6.54,
      "learning_rate": 5.498602275592873e-07,
      "loss": 0.053,
      "step": 6882,
      "task_loss": 0.01648840121924877
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993866645788987,
      "compression_loss": 0.0,
      "distillation_loss": 0.042155489325523376,
      "epoch": 6.54,
      "learning_rate": 5.476399913184438e-07,
      "loss": 0.0563,
      "step": 6883,
      "task_loss": 0.1832885593175888
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993878570648013,
      "compression_loss": 0.0,
      "distillation_loss": 0.01913582533597946,
      "epoch": 6.54,
      "learning_rate": 5.454241969665597e-07,
      "loss": 0.0177,
      "step": 6884,
      "task_loss": 0.005137601867318153
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993890480040297,
      "compression_loss": 0.0,
      "distillation_loss": 0.08212343603372574,
      "epoch": 6.54,
      "learning_rate": 5.432128449061464e-07,
      "loss": 0.085,
      "step": 6885,
      "task_loss": 0.11137494444847107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993902373975876,
      "compression_loss": 0.0,
      "distillation_loss": 0.034957289695739746,
      "epoch": 6.54,
      "learning_rate": 5.410059355388964e-07,
      "loss": 0.0336,
      "step": 6886,
      "task_loss": 0.020979827269911766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993914252464789,
      "compression_loss": 0.0,
      "distillation_loss": 0.026353981345891953,
      "epoch": 6.54,
      "learning_rate": 5.388034692657223e-07,
      "loss": 0.0242,
      "step": 6887,
      "task_loss": 0.004370139911770821
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799392611551707,
      "compression_loss": 0.0,
      "distillation_loss": 0.037230148911476135,
      "epoch": 6.54,
      "learning_rate": 5.366054464867016e-07,
      "loss": 0.0342,
      "step": 6888,
      "task_loss": 0.006543133407831192
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993937963142759,
      "compression_loss": 0.0,
      "distillation_loss": 0.019730228930711746,
      "epoch": 6.54,
      "learning_rate": 5.344118676011172e-07,
      "loss": 0.0349,
      "step": 6889,
      "task_loss": 0.17101413011550903
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799394979535189,
      "compression_loss": 0.0,
      "distillation_loss": 0.026037897914648056,
      "epoch": 6.54,
      "learning_rate": 5.322227330074481e-07,
      "loss": 0.0328,
      "step": 6890,
      "task_loss": 0.09383513033390045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993961612154501,
      "compression_loss": 0.0,
      "distillation_loss": 0.029655098915100098,
      "epoch": 6.54,
      "learning_rate": 5.300380431033564e-07,
      "loss": 0.0276,
      "step": 6891,
      "task_loss": 0.009574664756655693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993973413560629,
      "compression_loss": 0.0,
      "distillation_loss": 0.06856879591941833,
      "epoch": 6.55,
      "learning_rate": 5.278577982857025e-07,
      "loss": 0.0743,
      "step": 6892,
      "task_loss": 0.12565253674983978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799398519958031,
      "compression_loss": 0.0,
      "distillation_loss": 0.049631062895059586,
      "epoch": 6.55,
      "learning_rate": 5.25681998950539e-07,
      "loss": 0.046,
      "step": 6893,
      "task_loss": 0.013276774436235428
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7993996970223581,
      "compression_loss": 0.0,
      "distillation_loss": 0.05079222470521927,
      "epoch": 6.55,
      "learning_rate": 5.235106454931083e-07,
      "loss": 0.0519,
      "step": 6894,
      "task_loss": 0.06189654394984245
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799400872550048,
      "compression_loss": 0.0,
      "distillation_loss": 0.017339782789349556,
      "epoch": 6.55,
      "learning_rate": 5.213437383078501e-07,
      "loss": 0.0241,
      "step": 6895,
      "task_loss": 0.08445850759744644
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994020465421043,
      "compression_loss": 0.0,
      "distillation_loss": 0.08581778407096863,
      "epoch": 6.55,
      "learning_rate": 5.191812777883915e-07,
      "loss": 0.1178,
      "step": 6896,
      "task_loss": 0.40567508339881897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994032189995306,
      "compression_loss": 0.0,
      "distillation_loss": 0.02632337063550949,
      "epoch": 6.55,
      "learning_rate": 5.170232643275541e-07,
      "loss": 0.0352,
      "step": 6897,
      "task_loss": 0.11460480093955994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994043899233307,
      "compression_loss": 0.0,
      "distillation_loss": 0.09149836003780365,
      "epoch": 6.55,
      "learning_rate": 5.148696983173551e-07,
      "loss": 0.0872,
      "step": 6898,
      "task_loss": 0.048190876841545105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994055593145083,
      "compression_loss": 0.0,
      "distillation_loss": 0.08621042221784592,
      "epoch": 6.55,
      "learning_rate": 5.127205801489954e-07,
      "loss": 0.0784,
      "step": 6899,
      "task_loss": 0.0076434426009655
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799406727174067,
      "compression_loss": 0.0,
      "distillation_loss": 0.03059300407767296,
      "epoch": 6.55,
      "learning_rate": 5.105759102128738e-07,
      "loss": 0.0345,
      "step": 6900,
      "task_loss": 0.06991107016801834
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994078935030104,
      "compression_loss": 0.0,
      "distillation_loss": 0.01730835810303688,
      "epoch": 6.55,
      "learning_rate": 5.084356888985814e-07,
      "loss": 0.0217,
      "step": 6901,
      "task_loss": 0.06103728711605072
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994090583023424,
      "compression_loss": 0.0,
      "distillation_loss": 0.03369038552045822,
      "epoch": 6.55,
      "learning_rate": 5.062999165948989e-07,
      "loss": 0.04,
      "step": 6902,
      "task_loss": 0.09683724492788315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994102215730666,
      "compression_loss": 0.0,
      "distillation_loss": 0.057407524436712265,
      "epoch": 6.56,
      "learning_rate": 5.041685936897966e-07,
      "loss": 0.0604,
      "step": 6903,
      "task_loss": 0.08716972172260284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994113833161866,
      "compression_loss": 0.0,
      "distillation_loss": 0.06570513546466827,
      "epoch": 6.56,
      "learning_rate": 5.020417205704453e-07,
      "loss": 0.0642,
      "step": 6904,
      "task_loss": 0.0507512241601944
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994125435327062,
      "compression_loss": 0.0,
      "distillation_loss": 0.017435505986213684,
      "epoch": 6.56,
      "learning_rate": 4.999192976231998e-07,
      "loss": 0.026,
      "step": 6905,
      "task_loss": 0.10334107279777527
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799413702223629,
      "compression_loss": 0.0,
      "distillation_loss": 0.0625772476196289,
      "epoch": 6.56,
      "learning_rate": 4.978013252336072e-07,
      "loss": 0.0609,
      "step": 6906,
      "task_loss": 0.04585137218236923
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994148593899586,
      "compression_loss": 0.0,
      "distillation_loss": 0.07100007683038712,
      "epoch": 6.56,
      "learning_rate": 4.956878037864043e-07,
      "loss": 0.0801,
      "step": 6907,
      "task_loss": 0.1624176800251007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994160150326988,
      "compression_loss": 0.0,
      "distillation_loss": 0.0802350789308548,
      "epoch": 6.56,
      "learning_rate": 4.935787336655285e-07,
      "loss": 0.0845,
      "step": 6908,
      "task_loss": 0.12317005544900894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994171691528533,
      "compression_loss": 0.0,
      "distillation_loss": 0.024071305990219116,
      "epoch": 6.56,
      "learning_rate": 4.914741152541008e-07,
      "loss": 0.0222,
      "step": 6909,
      "task_loss": 0.005024924874305725
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994183217514258,
      "compression_loss": 0.0,
      "distillation_loss": 0.10285592824220657,
      "epoch": 6.56,
      "learning_rate": 4.893739489344323e-07,
      "loss": 0.1003,
      "step": 6910,
      "task_loss": 0.07742301374673843
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994194728294198,
      "compression_loss": 0.0,
      "distillation_loss": 0.0873519778251648,
      "epoch": 6.56,
      "learning_rate": 4.872782350880317e-07,
      "loss": 0.0973,
      "step": 6911,
      "task_loss": 0.18709777295589447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994206223878393,
      "compression_loss": 0.0,
      "distillation_loss": 0.02852596715092659,
      "epoch": 6.56,
      "learning_rate": 4.851869740955944e-07,
      "loss": 0.0342,
      "step": 6912,
      "task_loss": 0.08523042500019073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994217704276876,
      "compression_loss": 0.0,
      "distillation_loss": 0.15731999278068542,
      "epoch": 6.57,
      "learning_rate": 4.831001663370083e-07,
      "loss": 0.1699,
      "step": 6913,
      "task_loss": 0.2830032706260681
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994229169499687,
      "compression_loss": 0.0,
      "distillation_loss": 0.04235370457172394,
      "epoch": 6.57,
      "learning_rate": 4.810178121913478e-07,
      "loss": 0.0412,
      "step": 6914,
      "task_loss": 0.030333248898386955
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994240619556862,
      "compression_loss": 0.0,
      "distillation_loss": 0.07360881567001343,
      "epoch": 6.57,
      "learning_rate": 4.78939912036891e-07,
      "loss": 0.0884,
      "step": 6915,
      "task_loss": 0.22195890545845032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994252054458436,
      "compression_loss": 0.0,
      "distillation_loss": 0.017446476966142654,
      "epoch": 6.57,
      "learning_rate": 4.768664662510941e-07,
      "loss": 0.016,
      "step": 6916,
      "task_loss": 0.0031855106353759766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994263474214448,
      "compression_loss": 0.0,
      "distillation_loss": 0.023602165281772614,
      "epoch": 6.57,
      "learning_rate": 4.7479747521060324e-07,
      "loss": 0.0265,
      "step": 6917,
      "task_loss": 0.05219149589538574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994274878834934,
      "compression_loss": 0.0,
      "distillation_loss": 0.09219710528850555,
      "epoch": 6.57,
      "learning_rate": 4.727329392912705e-07,
      "loss": 0.0911,
      "step": 6918,
      "task_loss": 0.08144529163837433
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799428626832993,
      "compression_loss": 0.0,
      "distillation_loss": 0.08491960167884827,
      "epoch": 6.57,
      "learning_rate": 4.706728588681236e-07,
      "loss": 0.0811,
      "step": 6919,
      "task_loss": 0.04681730642914772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994297642709475,
      "compression_loss": 0.0,
      "distillation_loss": 0.10116090625524521,
      "epoch": 6.57,
      "learning_rate": 4.6861723431538276e-07,
      "loss": 0.0987,
      "step": 6920,
      "task_loss": 0.0767616331577301
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994309001983604,
      "compression_loss": 0.0,
      "distillation_loss": 0.012731763534247875,
      "epoch": 6.57,
      "learning_rate": 4.665660660064686e-07,
      "loss": 0.0123,
      "step": 6921,
      "task_loss": 0.008857103064656258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994320346162354,
      "compression_loss": 0.0,
      "distillation_loss": 0.02278589829802513,
      "epoch": 6.57,
      "learning_rate": 4.6451935431398306e-07,
      "loss": 0.021,
      "step": 6922,
      "task_loss": 0.004962848499417305
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994331675255762,
      "compression_loss": 0.0,
      "distillation_loss": 0.04454980418086052,
      "epoch": 6.57,
      "learning_rate": 4.6247709960972053e-07,
      "loss": 0.0438,
      "step": 6923,
      "task_loss": 0.03726205229759216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994342989273865,
      "compression_loss": 0.0,
      "distillation_loss": 0.07946009188890457,
      "epoch": 6.58,
      "learning_rate": 4.604393022646647e-07,
      "loss": 0.0817,
      "step": 6924,
      "task_loss": 0.10193922370672226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79943542882267,
      "compression_loss": 0.0,
      "distillation_loss": 0.03449847549200058,
      "epoch": 6.58,
      "learning_rate": 4.584059626489973e-07,
      "loss": 0.033,
      "step": 6925,
      "task_loss": 0.01942656934261322
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994365572124305,
      "compression_loss": 0.0,
      "distillation_loss": 0.027452457696199417,
      "epoch": 6.58,
      "learning_rate": 4.5637708113207834e-07,
      "loss": 0.0329,
      "step": 6926,
      "task_loss": 0.08229565620422363
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994376840976714,
      "compression_loss": 0.0,
      "distillation_loss": 0.014757486060261726,
      "epoch": 6.58,
      "learning_rate": 4.5435265808246585e-07,
      "loss": 0.0137,
      "step": 6927,
      "task_loss": 0.0037736501544713974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994388094793965,
      "compression_loss": 0.0,
      "distillation_loss": 0.07772176712751389,
      "epoch": 6.58,
      "learning_rate": 4.5233269386790734e-07,
      "loss": 0.0742,
      "step": 6928,
      "task_loss": 0.042066704481840134
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994399333586095,
      "compression_loss": 0.0,
      "distillation_loss": 0.08422841131687164,
      "epoch": 6.58,
      "learning_rate": 4.5031718885533423e-07,
      "loss": 0.083,
      "step": 6929,
      "task_loss": 0.07235755026340485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994410557363142,
      "compression_loss": 0.0,
      "distillation_loss": 0.04439088702201843,
      "epoch": 6.58,
      "learning_rate": 4.483061434108815e-07,
      "loss": 0.0417,
      "step": 6930,
      "task_loss": 0.017775850370526314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799442176613514,
      "compression_loss": 0.0,
      "distillation_loss": 0.03447749838232994,
      "epoch": 6.58,
      "learning_rate": 4.462995578998569e-07,
      "loss": 0.058,
      "step": 6931,
      "task_loss": 0.2692142724990845
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994432959912129,
      "compression_loss": 0.0,
      "distillation_loss": 0.06817720085382462,
      "epoch": 6.58,
      "learning_rate": 4.4429743268676884e-07,
      "loss": 0.0724,
      "step": 6932,
      "task_loss": 0.11086969822645187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994444138704144,
      "compression_loss": 0.0,
      "distillation_loss": 0.05889270454645157,
      "epoch": 6.58,
      "learning_rate": 4.4229976813531806e-07,
      "loss": 0.057,
      "step": 6933,
      "task_loss": 0.040453821420669556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994455302521222,
      "compression_loss": 0.0,
      "distillation_loss": 0.02058131992816925,
      "epoch": 6.58,
      "learning_rate": 4.403065646083809e-07,
      "loss": 0.019,
      "step": 6934,
      "task_loss": 0.00488920696079731
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79944664513734,
      "compression_loss": 0.0,
      "distillation_loss": 0.03927480801939964,
      "epoch": 6.59,
      "learning_rate": 4.383178224680401e-07,
      "loss": 0.0447,
      "step": 6935,
      "task_loss": 0.09392157196998596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994477585270715,
      "compression_loss": 0.0,
      "distillation_loss": 0.04294563829898834,
      "epoch": 6.59,
      "learning_rate": 4.3633354207555653e-07,
      "loss": 0.0536,
      "step": 6936,
      "task_loss": 0.1493779420852661
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994488704223204,
      "compression_loss": 0.0,
      "distillation_loss": 0.015475824475288391,
      "epoch": 6.59,
      "learning_rate": 4.3435372379138085e-07,
      "loss": 0.0143,
      "step": 6937,
      "task_loss": 0.0033792592585086823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994499808240902,
      "compression_loss": 0.0,
      "distillation_loss": 0.017580023035407066,
      "epoch": 6.59,
      "learning_rate": 4.3237836797516417e-07,
      "loss": 0.0163,
      "step": 6938,
      "task_loss": 0.004325071349740028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994510897333849,
      "compression_loss": 0.0,
      "distillation_loss": 0.024261541664600372,
      "epoch": 6.59,
      "learning_rate": 4.304074749857362e-07,
      "loss": 0.0256,
      "step": 6939,
      "task_loss": 0.037375908344984055
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799452197151208,
      "compression_loss": 0.0,
      "distillation_loss": 0.010520851239562035,
      "epoch": 6.59,
      "learning_rate": 4.284410451811188e-07,
      "loss": 0.0194,
      "step": 6940,
      "task_loss": 0.09917841851711273
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994533030785631,
      "compression_loss": 0.0,
      "distillation_loss": 0.04114697501063347,
      "epoch": 6.59,
      "learning_rate": 4.2647907891852357e-07,
      "loss": 0.0494,
      "step": 6941,
      "task_loss": 0.12371751666069031
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994544075164539,
      "compression_loss": 0.0,
      "distillation_loss": 0.02658040262758732,
      "epoch": 6.59,
      "learning_rate": 4.2452157655435145e-07,
      "loss": 0.0263,
      "step": 6942,
      "task_loss": 0.023485524579882622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994555104658843,
      "compression_loss": 0.0,
      "distillation_loss": 0.01725972816348076,
      "epoch": 6.59,
      "learning_rate": 4.225685384441902e-07,
      "loss": 0.0161,
      "step": 6943,
      "task_loss": 0.005400681868195534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994566119278578,
      "compression_loss": 0.0,
      "distillation_loss": 0.03645411878824234,
      "epoch": 6.59,
      "learning_rate": 4.2061996494282e-07,
      "loss": 0.0406,
      "step": 6944,
      "task_loss": 0.07785525172948837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994577119033781,
      "compression_loss": 0.0,
      "distillation_loss": 0.0437319241464138,
      "epoch": 6.6,
      "learning_rate": 4.1867585640421036e-07,
      "loss": 0.0513,
      "step": 6945,
      "task_loss": 0.11954252421855927
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994588103934489,
      "compression_loss": 0.0,
      "distillation_loss": 0.040995851159095764,
      "epoch": 6.6,
      "learning_rate": 4.16736213181515e-07,
      "loss": 0.0448,
      "step": 6946,
      "task_loss": 0.0789896547794342
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994599073990739,
      "compression_loss": 0.0,
      "distillation_loss": 0.0845775455236435,
      "epoch": 6.6,
      "learning_rate": 4.148010356270826e-07,
      "loss": 0.0876,
      "step": 6947,
      "task_loss": 0.11447571963071823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994610029212568,
      "compression_loss": 0.0,
      "distillation_loss": 0.019938606768846512,
      "epoch": 6.6,
      "learning_rate": 4.128703240924431e-07,
      "loss": 0.0185,
      "step": 6948,
      "task_loss": 0.00549742765724659
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994620969610012,
      "compression_loss": 0.0,
      "distillation_loss": 0.01663760282099247,
      "epoch": 6.6,
      "learning_rate": 4.109440789283242e-07,
      "loss": 0.0323,
      "step": 6949,
      "task_loss": 0.17326664924621582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994631895193108,
      "compression_loss": 0.0,
      "distillation_loss": 0.049342524260282516,
      "epoch": 6.6,
      "learning_rate": 4.0902230048463495e-07,
      "loss": 0.0508,
      "step": 6950,
      "task_loss": 0.0635058730840683
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994642805971894,
      "compression_loss": 0.0,
      "distillation_loss": 0.06009483337402344,
      "epoch": 6.6,
      "learning_rate": 4.071049891104739e-07,
      "loss": 0.0572,
      "step": 6951,
      "task_loss": 0.031034225597977638
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994653701956406,
      "compression_loss": 0.0,
      "distillation_loss": 0.03527653217315674,
      "epoch": 6.6,
      "learning_rate": 4.0519214515413463e-07,
      "loss": 0.0381,
      "step": 6952,
      "task_loss": 0.06350395828485489
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799466458315668,
      "compression_loss": 0.0,
      "distillation_loss": 0.056782066822052,
      "epoch": 6.6,
      "learning_rate": 4.0328376896309473e-07,
      "loss": 0.0651,
      "step": 6953,
      "task_loss": 0.13970398902893066
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994675449582754,
      "compression_loss": 0.0,
      "distillation_loss": 0.08646276593208313,
      "epoch": 6.6,
      "learning_rate": 4.01379860884013e-07,
      "loss": 0.0818,
      "step": 6954,
      "task_loss": 0.040067270398139954
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994686301244665,
      "compression_loss": 0.0,
      "distillation_loss": 0.08508466184139252,
      "epoch": 6.6,
      "learning_rate": 3.994804212627462e-07,
      "loss": 0.0789,
      "step": 6955,
      "task_loss": 0.02337714470922947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994697138152448,
      "compression_loss": 0.0,
      "distillation_loss": 0.04727325588464737,
      "epoch": 6.61,
      "learning_rate": 3.975854504443433e-07,
      "loss": 0.0431,
      "step": 6956,
      "task_loss": 0.005318284034729004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994707960316142,
      "compression_loss": 0.0,
      "distillation_loss": 0.025855958461761475,
      "epoch": 6.61,
      "learning_rate": 3.956949487730288e-07,
      "loss": 0.0347,
      "step": 6957,
      "task_loss": 0.11433203518390656
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994718767745782,
      "compression_loss": 0.0,
      "distillation_loss": 0.060321077704429626,
      "epoch": 6.61,
      "learning_rate": 3.9380891659221986e-07,
      "loss": 0.0617,
      "step": 6958,
      "task_loss": 0.07363105565309525
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994729560451407,
      "compression_loss": 0.0,
      "distillation_loss": 0.034057214856147766,
      "epoch": 6.61,
      "learning_rate": 3.9192735424452843e-07,
      "loss": 0.0396,
      "step": 6959,
      "task_loss": 0.08952410519123077
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994740338443052,
      "compression_loss": 0.0,
      "distillation_loss": 0.022792372852563858,
      "epoch": 6.61,
      "learning_rate": 3.900502620717478e-07,
      "loss": 0.0222,
      "step": 6960,
      "task_loss": 0.017220191657543182
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994751101730754,
      "compression_loss": 0.0,
      "distillation_loss": 0.02029343880712986,
      "epoch": 6.61,
      "learning_rate": 3.881776404148552e-07,
      "loss": 0.0242,
      "step": 6961,
      "task_loss": 0.059222202748060226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799476185032455,
      "compression_loss": 0.0,
      "distillation_loss": 0.026080617681145668,
      "epoch": 6.61,
      "learning_rate": 3.8630948961403125e-07,
      "loss": 0.0244,
      "step": 6962,
      "task_loss": 0.00916319526731968
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994772584234476,
      "compression_loss": 0.0,
      "distillation_loss": 0.02705274149775505,
      "epoch": 6.61,
      "learning_rate": 3.844458100086268e-07,
      "loss": 0.0329,
      "step": 6963,
      "task_loss": 0.08544865250587463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994783303470572,
      "compression_loss": 0.0,
      "distillation_loss": 0.02477836422622204,
      "epoch": 6.61,
      "learning_rate": 3.8258660193719044e-07,
      "loss": 0.0374,
      "step": 6964,
      "task_loss": 0.15073972940444946
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994794008042871,
      "compression_loss": 0.0,
      "distillation_loss": 0.022429198026657104,
      "epoch": 6.61,
      "learning_rate": 3.8073186573745757e-07,
      "loss": 0.0208,
      "step": 6965,
      "task_loss": 0.006342481821775436
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994804697961412,
      "compression_loss": 0.0,
      "distillation_loss": 0.08988398313522339,
      "epoch": 6.62,
      "learning_rate": 3.7888160174634757e-07,
      "loss": 0.0898,
      "step": 6966,
      "task_loss": 0.08916927129030228
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994815373236231,
      "compression_loss": 0.0,
      "distillation_loss": 0.028484515845775604,
      "epoch": 6.62,
      "learning_rate": 3.7703581029997215e-07,
      "loss": 0.0278,
      "step": 6967,
      "task_loss": 0.022118881344795227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994826033877366,
      "compression_loss": 0.0,
      "distillation_loss": 0.0302952341735363,
      "epoch": 6.62,
      "learning_rate": 3.75194491733627e-07,
      "loss": 0.028,
      "step": 6968,
      "task_loss": 0.0076552219688892365
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994836679894852,
      "compression_loss": 0.0,
      "distillation_loss": 0.03270921856164932,
      "epoch": 6.62,
      "learning_rate": 3.733576463817973e-07,
      "loss": 0.0308,
      "step": 6969,
      "task_loss": 0.013129375874996185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994847311298727,
      "compression_loss": 0.0,
      "distillation_loss": 0.052680712193250656,
      "epoch": 6.62,
      "learning_rate": 3.71525274578155e-07,
      "loss": 0.0622,
      "step": 6970,
      "task_loss": 0.1481049656867981
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994857928099028,
      "compression_loss": 0.0,
      "distillation_loss": 0.028156310319900513,
      "epoch": 6.62,
      "learning_rate": 3.696973766555589e-07,
      "loss": 0.0263,
      "step": 6971,
      "task_loss": 0.009480351582169533
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994868530305791,
      "compression_loss": 0.0,
      "distillation_loss": 0.036262333393096924,
      "epoch": 6.62,
      "learning_rate": 3.6787395294605455e-07,
      "loss": 0.0339,
      "step": 6972,
      "task_loss": 0.012746721506118774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994879117929053,
      "compression_loss": 0.0,
      "distillation_loss": 0.08903949707746506,
      "epoch": 6.62,
      "learning_rate": 3.660550037808741e-07,
      "loss": 0.1021,
      "step": 6973,
      "task_loss": 0.21985527873039246
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994889690978851,
      "compression_loss": 0.0,
      "distillation_loss": 0.03600417822599411,
      "epoch": 6.62,
      "learning_rate": 3.642405294904422e-07,
      "loss": 0.0337,
      "step": 6974,
      "task_loss": 0.01251133345067501
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994900249465222,
      "compression_loss": 0.0,
      "distillation_loss": 0.02271505631506443,
      "epoch": 6.62,
      "learning_rate": 3.624305304043646e-07,
      "loss": 0.0211,
      "step": 6975,
      "task_loss": 0.006729325279593468
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994910793398202,
      "compression_loss": 0.0,
      "distillation_loss": 0.04386524111032486,
      "epoch": 6.62,
      "learning_rate": 3.606250068514394e-07,
      "loss": 0.048,
      "step": 6976,
      "task_loss": 0.08489131927490234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799492132278783,
      "compression_loss": 0.0,
      "distillation_loss": 0.03540867939591408,
      "epoch": 6.63,
      "learning_rate": 3.5882395915964315e-07,
      "loss": 0.0324,
      "step": 6977,
      "task_loss": 0.005433699116110802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799493183764414,
      "compression_loss": 0.0,
      "distillation_loss": 0.04147842526435852,
      "epoch": 6.63,
      "learning_rate": 3.570273876561475e-07,
      "loss": 0.0387,
      "step": 6978,
      "task_loss": 0.013455674052238464
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994942337977171,
      "compression_loss": 0.0,
      "distillation_loss": 0.043742306530475616,
      "epoch": 6.63,
      "learning_rate": 3.552352926673136e-07,
      "loss": 0.0518,
      "step": 6979,
      "task_loss": 0.12435872107744217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994952823796958,
      "compression_loss": 0.0,
      "distillation_loss": 0.04596320539712906,
      "epoch": 6.63,
      "learning_rate": 3.5344767451867545e-07,
      "loss": 0.0448,
      "step": 6980,
      "task_loss": 0.03420111536979675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799496329511354,
      "compression_loss": 0.0,
      "distillation_loss": 0.09754104167222977,
      "epoch": 6.63,
      "learning_rate": 3.51664533534965e-07,
      "loss": 0.1004,
      "step": 6981,
      "task_loss": 0.1259613186120987
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994973751936951,
      "compression_loss": 0.0,
      "distillation_loss": 0.014837587252259254,
      "epoch": 6.63,
      "learning_rate": 3.498858700401009e-07,
      "loss": 0.0224,
      "step": 6982,
      "task_loss": 0.09078174829483032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799498419427723,
      "compression_loss": 0.0,
      "distillation_loss": 0.07896040380001068,
      "epoch": 6.63,
      "learning_rate": 3.481116843571858e-07,
      "loss": 0.0744,
      "step": 6983,
      "task_loss": 0.03332065790891647
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7994994622144413,
      "compression_loss": 0.0,
      "distillation_loss": 0.05292394757270813,
      "epoch": 6.63,
      "learning_rate": 3.463419768085091e-07,
      "loss": 0.0551,
      "step": 6984,
      "task_loss": 0.07501198351383209
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995005035548538,
      "compression_loss": 0.0,
      "distillation_loss": 0.019998980686068535,
      "epoch": 6.63,
      "learning_rate": 3.4457674771554425e-07,
      "loss": 0.0186,
      "step": 6985,
      "task_loss": 0.005534190684556961
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799501543449964,
      "compression_loss": 0.0,
      "distillation_loss": 0.03253600001335144,
      "epoch": 6.63,
      "learning_rate": 3.428159973989542e-07,
      "loss": 0.0317,
      "step": 6986,
      "task_loss": 0.024117592722177505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995025819007757,
      "compression_loss": 0.0,
      "distillation_loss": 0.02256142720580101,
      "epoch": 6.64,
      "learning_rate": 3.4105972617859136e-07,
      "loss": 0.0233,
      "step": 6987,
      "task_loss": 0.029686935245990753
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995036189082926,
      "compression_loss": 0.0,
      "distillation_loss": 0.02714642882347107,
      "epoch": 6.64,
      "learning_rate": 3.3930793437348675e-07,
      "loss": 0.0258,
      "step": 6988,
      "task_loss": 0.013444697484374046
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995046544735183,
      "compression_loss": 0.0,
      "distillation_loss": 0.08077394962310791,
      "epoch": 6.64,
      "learning_rate": 3.3756062230186067e-07,
      "loss": 0.0878,
      "step": 6989,
      "task_loss": 0.15142491459846497
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995056885974564,
      "compression_loss": 0.0,
      "distillation_loss": 0.10645799338817596,
      "epoch": 6.64,
      "learning_rate": 3.358177902811233e-07,
      "loss": 0.1093,
      "step": 6990,
      "task_loss": 0.134621262550354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995067212811109,
      "compression_loss": 0.0,
      "distillation_loss": 0.020988423377275467,
      "epoch": 6.64,
      "learning_rate": 3.340794386278712e-07,
      "loss": 0.0284,
      "step": 6991,
      "task_loss": 0.094853974878788
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995077525254851,
      "compression_loss": 0.0,
      "distillation_loss": 0.05726141855120659,
      "epoch": 6.64,
      "learning_rate": 3.3234556765787963e-07,
      "loss": 0.0633,
      "step": 6992,
      "task_loss": 0.1174088567495346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799508782331583,
      "compression_loss": 0.0,
      "distillation_loss": 0.05345267802476883,
      "epoch": 6.64,
      "learning_rate": 3.306161776861161e-07,
      "loss": 0.0579,
      "step": 6993,
      "task_loss": 0.0983385294675827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799509810700408,
      "compression_loss": 0.0,
      "distillation_loss": 0.0895034670829773,
      "epoch": 6.64,
      "learning_rate": 3.288912690267348e-07,
      "loss": 0.0974,
      "step": 6994,
      "task_loss": 0.16841556131839752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799510837632964,
      "compression_loss": 0.0,
      "distillation_loss": 0.020515451207756996,
      "epoch": 6.64,
      "learning_rate": 3.2717084199307134e-07,
      "loss": 0.0191,
      "step": 6995,
      "task_loss": 0.006037186831235886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995118631302546,
      "compression_loss": 0.0,
      "distillation_loss": 0.02419309876859188,
      "epoch": 6.64,
      "learning_rate": 3.2545489689764784e-07,
      "loss": 0.0414,
      "step": 6996,
      "task_loss": 0.19664430618286133
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995128871932835,
      "compression_loss": 0.0,
      "distillation_loss": 0.0171881765127182,
      "epoch": 6.64,
      "learning_rate": 3.237434340521789e-07,
      "loss": 0.0199,
      "step": 6997,
      "task_loss": 0.044096946716308594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995139098230544,
      "compression_loss": 0.0,
      "distillation_loss": 0.07256430387496948,
      "epoch": 6.65,
      "learning_rate": 3.220364537675574e-07,
      "loss": 0.0694,
      "step": 6998,
      "task_loss": 0.04073936119675636
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995149310205709,
      "compression_loss": 0.0,
      "distillation_loss": 0.07214680314064026,
      "epoch": 6.65,
      "learning_rate": 3.203339563538632e-07,
      "loss": 0.067,
      "step": 6999,
      "task_loss": 0.021014513447880745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995159507868367,
      "compression_loss": 0.0,
      "distillation_loss": 0.03768681734800339,
      "epoch": 6.65,
      "learning_rate": 3.1863594212036274e-07,
      "loss": 0.0393,
      "step": 7000,
      "task_loss": 0.053721100091934204
    },
    {
      "epoch": 6.65,
      "eval_accuracy": 0.8922018348623854,
      "eval_loss": 0.4350743889808655,
      "eval_runtime": 18.1267,
      "eval_samples_per_second": 48.106,
      "eval_steps_per_second": 6.013,
      "step": 7000
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995169691228556,
      "compression_loss": 0.0,
      "distillation_loss": 0.049745362251996994,
      "epoch": 6.65,
      "learning_rate": 3.1694241137551203e-07,
      "loss": 0.0466,
      "step": 7001,
      "task_loss": 0.018015502020716667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995179860296312,
      "compression_loss": 0.0,
      "distillation_loss": 0.029364963993430138,
      "epoch": 6.65,
      "learning_rate": 3.1525336442694843e-07,
      "loss": 0.0338,
      "step": 7002,
      "task_loss": 0.07397031784057617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995190015081671,
      "compression_loss": 0.0,
      "distillation_loss": 0.02187550999224186,
      "epoch": 6.65,
      "learning_rate": 3.1356880158149025e-07,
      "loss": 0.0202,
      "step": 7003,
      "task_loss": 0.0046534184366464615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995200155594672,
      "compression_loss": 0.0,
      "distillation_loss": 0.013252021744847298,
      "epoch": 6.65,
      "learning_rate": 3.118887231451539e-07,
      "loss": 0.0126,
      "step": 7004,
      "task_loss": 0.006522929295897484
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995210281845349,
      "compression_loss": 0.0,
      "distillation_loss": 0.039753131568431854,
      "epoch": 6.65,
      "learning_rate": 3.1021312942313144e-07,
      "loss": 0.0451,
      "step": 7005,
      "task_loss": 0.09337079524993896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799522039384374,
      "compression_loss": 0.0,
      "distillation_loss": 0.11256477236747742,
      "epoch": 6.65,
      "learning_rate": 3.0854202071979865e-07,
      "loss": 0.1108,
      "step": 7006,
      "task_loss": 0.09483619034290314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995230491599884,
      "compression_loss": 0.0,
      "distillation_loss": 0.024704527109861374,
      "epoch": 6.65,
      "learning_rate": 3.0687539733872115e-07,
      "loss": 0.0228,
      "step": 7007,
      "task_loss": 0.006121648475527763
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995240575123814,
      "compression_loss": 0.0,
      "distillation_loss": 0.030473047867417336,
      "epoch": 6.66,
      "learning_rate": 3.05213259582654e-07,
      "loss": 0.0289,
      "step": 7008,
      "task_loss": 0.015124401077628136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799525064442557,
      "compression_loss": 0.0,
      "distillation_loss": 0.021761227399110794,
      "epoch": 6.66,
      "learning_rate": 3.035556077535306e-07,
      "loss": 0.0294,
      "step": 7009,
      "task_loss": 0.09853780269622803
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995260699515188,
      "compression_loss": 0.0,
      "distillation_loss": 0.03172404319047928,
      "epoch": 6.66,
      "learning_rate": 3.0190244215246857e-07,
      "loss": 0.0289,
      "step": 7010,
      "task_loss": 0.003925886005163193
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995270740402703,
      "compression_loss": 0.0,
      "distillation_loss": 0.062315717339515686,
      "epoch": 6.66,
      "learning_rate": 3.002537630797747e-07,
      "loss": 0.0584,
      "step": 7011,
      "task_loss": 0.02332870475947857
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995280767098154,
      "compression_loss": 0.0,
      "distillation_loss": 0.03596599027514458,
      "epoch": 6.66,
      "learning_rate": 2.986095708349429e-07,
      "loss": 0.034,
      "step": 7012,
      "task_loss": 0.01661401055753231
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995290779611577,
      "compression_loss": 0.0,
      "distillation_loss": 0.03480362519621849,
      "epoch": 6.66,
      "learning_rate": 2.9696986571664253e-07,
      "loss": 0.0327,
      "step": 7013,
      "task_loss": 0.013543521985411644
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995300777953008,
      "compression_loss": 0.0,
      "distillation_loss": 0.024172749370336533,
      "epoch": 6.66,
      "learning_rate": 2.953346480227409e-07,
      "loss": 0.0245,
      "step": 7014,
      "task_loss": 0.027264190837740898
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995310762132486,
      "compression_loss": 0.0,
      "distillation_loss": 0.02965361624956131,
      "epoch": 6.66,
      "learning_rate": 2.937039180502782e-07,
      "loss": 0.0448,
      "step": 7015,
      "task_loss": 0.18084561824798584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995320732160046,
      "compression_loss": 0.0,
      "distillation_loss": 0.0646897479891777,
      "epoch": 6.66,
      "learning_rate": 2.920776760954813e-07,
      "loss": 0.0833,
      "step": 7016,
      "task_loss": 0.2510213255882263
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995330688045724,
      "compression_loss": 0.0,
      "distillation_loss": 0.03607410565018654,
      "epoch": 6.66,
      "learning_rate": 2.904559224537723e-07,
      "loss": 0.0336,
      "step": 7017,
      "task_loss": 0.011309439316391945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799534062979956,
      "compression_loss": 0.0,
      "distillation_loss": 0.0305813979357481,
      "epoch": 6.66,
      "learning_rate": 2.888386574197488e-07,
      "loss": 0.0356,
      "step": 7018,
      "task_loss": 0.08079958707094193
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995350557431589,
      "compression_loss": 0.0,
      "distillation_loss": 0.022271297872066498,
      "epoch": 6.67,
      "learning_rate": 2.872258812871925e-07,
      "loss": 0.0463,
      "step": 7019,
      "task_loss": 0.26298779249191284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995360470951847,
      "compression_loss": 0.0,
      "distillation_loss": 0.03892994672060013,
      "epoch": 6.67,
      "learning_rate": 2.8561759434907185e-07,
      "loss": 0.0363,
      "step": 7020,
      "task_loss": 0.012905815616250038
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799537037037037,
      "compression_loss": 0.0,
      "distillation_loss": 0.02603466808795929,
      "epoch": 6.67,
      "learning_rate": 2.84013796897542e-07,
      "loss": 0.0408,
      "step": 7021,
      "task_loss": 0.17337138950824738
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995380255697199,
      "compression_loss": 0.0,
      "distillation_loss": 0.024577513337135315,
      "epoch": 6.67,
      "learning_rate": 2.8241448922393945e-07,
      "loss": 0.0225,
      "step": 7022,
      "task_loss": 0.003398273140192032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995390126942367,
      "compression_loss": 0.0,
      "distillation_loss": 0.048275694251060486,
      "epoch": 6.67,
      "learning_rate": 2.8081967161878443e-07,
      "loss": 0.053,
      "step": 7023,
      "task_loss": 0.09526211768388748
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995399984115912,
      "compression_loss": 0.0,
      "distillation_loss": 0.029721971601247787,
      "epoch": 6.67,
      "learning_rate": 2.7922934437178695e-07,
      "loss": 0.043,
      "step": 7024,
      "task_loss": 0.16234955191612244
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995409827227872,
      "compression_loss": 0.0,
      "distillation_loss": 0.04240930825471878,
      "epoch": 6.67,
      "learning_rate": 2.7764350777183533e-07,
      "loss": 0.0402,
      "step": 7025,
      "task_loss": 0.02065758965909481
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995419656288282,
      "compression_loss": 0.0,
      "distillation_loss": 0.08198815584182739,
      "epoch": 6.67,
      "learning_rate": 2.760621621070047e-07,
      "loss": 0.0865,
      "step": 7026,
      "task_loss": 0.126779243350029
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995429471307178,
      "compression_loss": 0.0,
      "distillation_loss": 0.022945459932088852,
      "epoch": 6.67,
      "learning_rate": 2.744853076645515e-07,
      "loss": 0.0216,
      "step": 7027,
      "task_loss": 0.009481718763709068
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79954392722946,
      "compression_loss": 0.0,
      "distillation_loss": 0.0208904929459095,
      "epoch": 6.67,
      "learning_rate": 2.729129447309242e-07,
      "loss": 0.0194,
      "step": 7028,
      "task_loss": 0.006270444020628929
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995449059260583,
      "compression_loss": 0.0,
      "distillation_loss": 0.05013927444815636,
      "epoch": 6.68,
      "learning_rate": 2.713450735917472e-07,
      "loss": 0.0459,
      "step": 7029,
      "task_loss": 0.007837006822228432
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995458832215164,
      "compression_loss": 0.0,
      "distillation_loss": 0.04295579344034195,
      "epoch": 6.68,
      "learning_rate": 2.6978169453183153e-07,
      "loss": 0.0415,
      "step": 7030,
      "task_loss": 0.028465719893574715
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799546859116838,
      "compression_loss": 0.0,
      "distillation_loss": 0.03028765879571438,
      "epoch": 6.68,
      "learning_rate": 2.6822280783517504e-07,
      "loss": 0.0279,
      "step": 7031,
      "task_loss": 0.006683392450213432
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995478336130267,
      "compression_loss": 0.0,
      "distillation_loss": 0.025525229051709175,
      "epoch": 6.68,
      "learning_rate": 2.66668413784954e-07,
      "loss": 0.0238,
      "step": 7032,
      "task_loss": 0.008154524490237236
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995488067110863,
      "compression_loss": 0.0,
      "distillation_loss": 0.020436950027942657,
      "epoch": 6.68,
      "learning_rate": 2.651185126635314e-07,
      "loss": 0.0213,
      "step": 7033,
      "task_loss": 0.029449906200170517
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995497784120204,
      "compression_loss": 0.0,
      "distillation_loss": 0.020888283848762512,
      "epoch": 6.68,
      "learning_rate": 2.635731047524542e-07,
      "loss": 0.0276,
      "step": 7034,
      "task_loss": 0.08772215992212296
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995507487168327,
      "compression_loss": 0.0,
      "distillation_loss": 0.057622358202934265,
      "epoch": 6.68,
      "learning_rate": 2.620321903324563e-07,
      "loss": 0.0553,
      "step": 7035,
      "task_loss": 0.034089043736457825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995517176265269,
      "compression_loss": 0.0,
      "distillation_loss": 0.021918507292866707,
      "epoch": 6.68,
      "learning_rate": 2.6049576968345256e-07,
      "loss": 0.0204,
      "step": 7036,
      "task_loss": 0.006641771644353867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995526851421066,
      "compression_loss": 0.0,
      "distillation_loss": 0.01303351204842329,
      "epoch": 6.68,
      "learning_rate": 2.589638430845337e-07,
      "loss": 0.0124,
      "step": 7037,
      "task_loss": 0.00620032474398613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995536512645757,
      "compression_loss": 0.0,
      "distillation_loss": 0.04416292533278465,
      "epoch": 6.68,
      "learning_rate": 2.5743641081399094e-07,
      "loss": 0.0467,
      "step": 7038,
      "task_loss": 0.0692160353064537
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995546159949376,
      "compression_loss": 0.0,
      "distillation_loss": 0.044526465237140656,
      "epoch": 6.68,
      "learning_rate": 2.559134731492857e-07,
      "loss": 0.0502,
      "step": 7039,
      "task_loss": 0.10125073045492172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995555793341962,
      "compression_loss": 0.0,
      "distillation_loss": 0.06591819226741791,
      "epoch": 6.69,
      "learning_rate": 2.5439503036706615e-07,
      "loss": 0.0642,
      "step": 7040,
      "task_loss": 0.04842384532094002
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995565412833551,
      "compression_loss": 0.0,
      "distillation_loss": 0.04874798282980919,
      "epoch": 6.69,
      "learning_rate": 2.5288108274316435e-07,
      "loss": 0.0493,
      "step": 7041,
      "task_loss": 0.05393533781170845
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799557501843418,
      "compression_loss": 0.0,
      "distillation_loss": 0.05359012633562088,
      "epoch": 6.69,
      "learning_rate": 2.5137163055259926e-07,
      "loss": 0.0642,
      "step": 7042,
      "task_loss": 0.15967698395252228
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995584610153885,
      "compression_loss": 0.0,
      "distillation_loss": 0.20667728781700134,
      "epoch": 6.69,
      "learning_rate": 2.4986667406956544e-07,
      "loss": 0.2091,
      "step": 7043,
      "task_loss": 0.23065844178199768
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995594188002704,
      "compression_loss": 0.0,
      "distillation_loss": 0.06716637313365936,
      "epoch": 6.69,
      "learning_rate": 2.4836621356744973e-07,
      "loss": 0.0693,
      "step": 7044,
      "task_loss": 0.08879762887954712
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995603751990673,
      "compression_loss": 0.0,
      "distillation_loss": 0.07034559547901154,
      "epoch": 6.69,
      "learning_rate": 2.468702493188174e-07,
      "loss": 0.0752,
      "step": 7045,
      "task_loss": 0.1193457618355751
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799561330212783,
      "compression_loss": 0.0,
      "distillation_loss": 0.02908926270902157,
      "epoch": 6.69,
      "learning_rate": 2.453787815954178e-07,
      "loss": 0.0302,
      "step": 7046,
      "task_loss": 0.03969724103808403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995622838424209,
      "compression_loss": 0.0,
      "distillation_loss": 0.07951683551073074,
      "epoch": 6.69,
      "learning_rate": 2.4389181066817865e-07,
      "loss": 0.0891,
      "step": 7047,
      "task_loss": 0.17507222294807434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799563236088985,
      "compression_loss": 0.0,
      "distillation_loss": 0.009979777969419956,
      "epoch": 6.69,
      "learning_rate": 2.4240933680722e-07,
      "loss": 0.0095,
      "step": 7048,
      "task_loss": 0.004972290247678757
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995641869534789,
      "compression_loss": 0.0,
      "distillation_loss": 0.014456565491855145,
      "epoch": 6.69,
      "learning_rate": 2.4093136028184024e-07,
      "loss": 0.0133,
      "step": 7049,
      "task_loss": 0.003056548535823822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995651364369062,
      "compression_loss": 0.0,
      "distillation_loss": 0.06412947177886963,
      "epoch": 6.7,
      "learning_rate": 2.394578813605192e-07,
      "loss": 0.0856,
      "step": 7050,
      "task_loss": 0.2784784734249115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995660845402706,
      "compression_loss": 0.0,
      "distillation_loss": 0.048369936645030975,
      "epoch": 6.7,
      "learning_rate": 2.3798890031092037e-07,
      "loss": 0.0595,
      "step": 7051,
      "task_loss": 0.15925872325897217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995670312645757,
      "compression_loss": 0.0,
      "distillation_loss": 0.039365239441394806,
      "epoch": 6.7,
      "learning_rate": 2.3652441739989427e-07,
      "loss": 0.0365,
      "step": 7052,
      "task_loss": 0.010832777246832848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995679766108255,
      "compression_loss": 0.0,
      "distillation_loss": 0.08087919652462006,
      "epoch": 6.7,
      "learning_rate": 2.3506443289347259e-07,
      "loss": 0.0776,
      "step": 7053,
      "task_loss": 0.04851894453167915
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995689205800234,
      "compression_loss": 0.0,
      "distillation_loss": 0.09845395386219025,
      "epoch": 6.7,
      "learning_rate": 2.3360894705686254e-07,
      "loss": 0.1004,
      "step": 7054,
      "task_loss": 0.11757258325815201
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799569863173173,
      "compression_loss": 0.0,
      "distillation_loss": 0.020045258104801178,
      "epoch": 6.7,
      "learning_rate": 2.3215796015446378e-07,
      "loss": 0.0187,
      "step": 7055,
      "task_loss": 0.006300348788499832
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995708043912784,
      "compression_loss": 0.0,
      "distillation_loss": 0.06473547220230103,
      "epoch": 6.7,
      "learning_rate": 2.3071147244985713e-07,
      "loss": 0.0787,
      "step": 7056,
      "task_loss": 0.2046293467283249
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995717442353428,
      "compression_loss": 0.0,
      "distillation_loss": 0.08796077966690063,
      "epoch": 6.7,
      "learning_rate": 2.292694842057991e-07,
      "loss": 0.1015,
      "step": 7057,
      "task_loss": 0.22374670207500458
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995726827063702,
      "compression_loss": 0.0,
      "distillation_loss": 0.021141093224287033,
      "epoch": 6.7,
      "learning_rate": 2.278319956842384e-07,
      "loss": 0.0362,
      "step": 7058,
      "task_loss": 0.17210184037685394
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995736198053641,
      "compression_loss": 0.0,
      "distillation_loss": 0.03681035712361336,
      "epoch": 6.7,
      "learning_rate": 2.263990071462968e-07,
      "loss": 0.0371,
      "step": 7059,
      "task_loss": 0.039506033062934875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995745555333283,
      "compression_loss": 0.0,
      "distillation_loss": 0.03608547896146774,
      "epoch": 6.7,
      "learning_rate": 2.2497051885228827e-07,
      "loss": 0.0381,
      "step": 7060,
      "task_loss": 0.05643710866570473
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995754898912665,
      "compression_loss": 0.0,
      "distillation_loss": 0.015874043107032776,
      "epoch": 6.71,
      "learning_rate": 2.2354653106170244e-07,
      "loss": 0.015,
      "step": 7061,
      "task_loss": 0.006745463237166405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995764228801822,
      "compression_loss": 0.0,
      "distillation_loss": 0.03536041080951691,
      "epoch": 6.71,
      "learning_rate": 2.2212704403321572e-07,
      "loss": 0.0436,
      "step": 7062,
      "task_loss": 0.11737397313117981
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995773545010793,
      "compression_loss": 0.0,
      "distillation_loss": 0.20444419980049133,
      "epoch": 6.71,
      "learning_rate": 2.2071205802468299e-07,
      "loss": 0.194,
      "step": 7063,
      "task_loss": 0.10004756599664688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995782847549614,
      "compression_loss": 0.0,
      "distillation_loss": 0.04743964970111847,
      "epoch": 6.71,
      "learning_rate": 2.1930157329314026e-07,
      "loss": 0.0468,
      "step": 7064,
      "task_loss": 0.041466888040304184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995792136428321,
      "compression_loss": 0.0,
      "distillation_loss": 0.021672185510396957,
      "epoch": 6.71,
      "learning_rate": 2.178955900948132e-07,
      "loss": 0.0216,
      "step": 7065,
      "task_loss": 0.0210067518055439
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995801411656952,
      "compression_loss": 0.0,
      "distillation_loss": 0.04924159497022629,
      "epoch": 6.71,
      "learning_rate": 2.1649410868510577e-07,
      "loss": 0.05,
      "step": 7066,
      "task_loss": 0.056573107838630676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995810673245544,
      "compression_loss": 0.0,
      "distillation_loss": 0.03908860683441162,
      "epoch": 6.71,
      "learning_rate": 2.1509712931860327e-07,
      "loss": 0.0441,
      "step": 7067,
      "task_loss": 0.08947930485010147
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995819921204131,
      "compression_loss": 0.0,
      "distillation_loss": 0.140781432390213,
      "epoch": 6.71,
      "learning_rate": 2.1370465224906655e-07,
      "loss": 0.1488,
      "step": 7068,
      "task_loss": 0.2205652892589569
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995829155542754,
      "compression_loss": 0.0,
      "distillation_loss": 0.06154758110642433,
      "epoch": 6.71,
      "learning_rate": 2.1231667772945718e-07,
      "loss": 0.0578,
      "step": 7069,
      "task_loss": 0.02428608387708664
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995838376271447,
      "compression_loss": 0.0,
      "distillation_loss": 0.08523669838905334,
      "epoch": 6.71,
      "learning_rate": 2.1093320601190124e-07,
      "loss": 0.0834,
      "step": 7070,
      "task_loss": 0.06637626141309738
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995847583400248,
      "compression_loss": 0.0,
      "distillation_loss": 0.058525826781988144,
      "epoch": 6.72,
      "learning_rate": 2.0955423734771162e-07,
      "loss": 0.0756,
      "step": 7071,
      "task_loss": 0.2295752465724945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995856776939194,
      "compression_loss": 0.0,
      "distillation_loss": 0.03819255903363228,
      "epoch": 6.72,
      "learning_rate": 2.081797719873879e-07,
      "loss": 0.0503,
      "step": 7072,
      "task_loss": 0.15918831527233124
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799586595689832,
      "compression_loss": 0.0,
      "distillation_loss": 0.12596416473388672,
      "epoch": 6.72,
      "learning_rate": 2.0680981018060819e-07,
      "loss": 0.1229,
      "step": 7073,
      "task_loss": 0.09560129046440125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995875123287665,
      "compression_loss": 0.0,
      "distillation_loss": 0.026970701292157173,
      "epoch": 6.72,
      "learning_rate": 2.054443521762317e-07,
      "loss": 0.038,
      "step": 7074,
      "task_loss": 0.13767385482788086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995884276117264,
      "compression_loss": 0.0,
      "distillation_loss": 0.017990119755268097,
      "epoch": 6.72,
      "learning_rate": 2.040833982223017e-07,
      "loss": 0.0198,
      "step": 7075,
      "task_loss": 0.036511633545160294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995893415397156,
      "compression_loss": 0.0,
      "distillation_loss": 0.040901899337768555,
      "epoch": 6.72,
      "learning_rate": 2.0272694856603991e-07,
      "loss": 0.0423,
      "step": 7076,
      "task_loss": 0.05451485887169838
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995902541137376,
      "compression_loss": 0.0,
      "distillation_loss": 0.04028903320431709,
      "epoch": 6.72,
      "learning_rate": 2.013750034538492e-07,
      "loss": 0.0384,
      "step": 7077,
      "task_loss": 0.021871501579880714
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995911653347961,
      "compression_loss": 0.0,
      "distillation_loss": 0.020981252193450928,
      "epoch": 6.72,
      "learning_rate": 2.0002756313132475e-07,
      "loss": 0.0195,
      "step": 7078,
      "task_loss": 0.006563212722539902
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995920752038949,
      "compression_loss": 0.0,
      "distillation_loss": 0.0342644639313221,
      "epoch": 6.72,
      "learning_rate": 1.9868462784322917e-07,
      "loss": 0.036,
      "step": 7079,
      "task_loss": 0.05136824771761894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995929837220376,
      "compression_loss": 0.0,
      "distillation_loss": 0.1288357377052307,
      "epoch": 6.72,
      "learning_rate": 1.9734619783351727e-07,
      "loss": 0.1282,
      "step": 7080,
      "task_loss": 0.12262624502182007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995938908902278,
      "compression_loss": 0.0,
      "distillation_loss": 0.07863984256982803,
      "epoch": 6.72,
      "learning_rate": 1.9601227334531956e-07,
      "loss": 0.0845,
      "step": 7081,
      "task_loss": 0.13708055019378662
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995947967094693,
      "compression_loss": 0.0,
      "distillation_loss": 0.04373558238148689,
      "epoch": 6.73,
      "learning_rate": 1.946828546209478e-07,
      "loss": 0.0465,
      "step": 7082,
      "task_loss": 0.07146552950143814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995957011807658,
      "compression_loss": 0.0,
      "distillation_loss": 0.07881873846054077,
      "epoch": 6.73,
      "learning_rate": 1.9335794190190327e-07,
      "loss": 0.0787,
      "step": 7083,
      "task_loss": 0.0776781439781189
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995966043051209,
      "compression_loss": 0.0,
      "distillation_loss": 0.023989124223589897,
      "epoch": 6.73,
      "learning_rate": 1.920375354288545e-07,
      "loss": 0.0223,
      "step": 7084,
      "task_loss": 0.006650885567069054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995975060835383,
      "compression_loss": 0.0,
      "distillation_loss": 0.01962425746023655,
      "epoch": 6.73,
      "learning_rate": 1.9072163544166244e-07,
      "loss": 0.0248,
      "step": 7085,
      "task_loss": 0.07158499956130981
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995984065170217,
      "compression_loss": 0.0,
      "distillation_loss": 0.029367784038186073,
      "epoch": 6.73,
      "learning_rate": 1.89410242179372e-07,
      "loss": 0.0391,
      "step": 7086,
      "task_loss": 0.12671338021755219
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7995993056065748,
      "compression_loss": 0.0,
      "distillation_loss": 0.08097614347934723,
      "epoch": 6.73,
      "learning_rate": 1.881033558802009e-07,
      "loss": 0.0883,
      "step": 7087,
      "task_loss": 0.15451526641845703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996002033532013,
      "compression_loss": 0.0,
      "distillation_loss": 0.09910960495471954,
      "epoch": 6.73,
      "learning_rate": 1.8680097678154817e-07,
      "loss": 0.1072,
      "step": 7088,
      "task_loss": 0.17986391484737396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996010997579047,
      "compression_loss": 0.0,
      "distillation_loss": 0.03139527514576912,
      "epoch": 6.73,
      "learning_rate": 1.8550310511999958e-07,
      "loss": 0.0354,
      "step": 7089,
      "task_loss": 0.0717248022556305
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996019948216889,
      "compression_loss": 0.0,
      "distillation_loss": 0.020029647275805473,
      "epoch": 6.73,
      "learning_rate": 1.8420974113131927e-07,
      "loss": 0.0187,
      "step": 7090,
      "task_loss": 0.006666556000709534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996028885455575,
      "compression_loss": 0.0,
      "distillation_loss": 0.03404942527413368,
      "epoch": 6.73,
      "learning_rate": 1.8292088505045546e-07,
      "loss": 0.0398,
      "step": 7091,
      "task_loss": 0.09123058617115021
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996037809305142,
      "compression_loss": 0.0,
      "distillation_loss": 0.03140270709991455,
      "epoch": 6.74,
      "learning_rate": 1.8163653711153194e-07,
      "loss": 0.0426,
      "step": 7092,
      "task_loss": 0.14372628927230835
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996046719775627,
      "compression_loss": 0.0,
      "distillation_loss": 0.027259420603513718,
      "epoch": 6.74,
      "learning_rate": 1.8035669754785934e-07,
      "loss": 0.0421,
      "step": 7093,
      "task_loss": 0.17557671666145325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996055616877066,
      "compression_loss": 0.0,
      "distillation_loss": 0.023833533748984337,
      "epoch": 6.74,
      "learning_rate": 1.7908136659192387e-07,
      "loss": 0.0223,
      "step": 7094,
      "task_loss": 0.00828717090189457
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996064500619496,
      "compression_loss": 0.0,
      "distillation_loss": 0.11141105741262436,
      "epoch": 6.74,
      "learning_rate": 1.7781054447539858e-07,
      "loss": 0.1124,
      "step": 7095,
      "task_loss": 0.12159767001867294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996073371012955,
      "compression_loss": 0.0,
      "distillation_loss": 0.0840909332036972,
      "epoch": 6.74,
      "learning_rate": 1.7654423142913213e-07,
      "loss": 0.0868,
      "step": 7096,
      "task_loss": 0.1110580712556839
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996082228067479,
      "compression_loss": 0.0,
      "distillation_loss": 0.018214348703622818,
      "epoch": 6.74,
      "learning_rate": 1.7528242768315717e-07,
      "loss": 0.0171,
      "step": 7097,
      "task_loss": 0.007498534396290779
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996091071793104,
      "compression_loss": 0.0,
      "distillation_loss": 0.02279078960418701,
      "epoch": 6.74,
      "learning_rate": 1.7402513346668758e-07,
      "loss": 0.021,
      "step": 7098,
      "task_loss": 0.0046605877578258514
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996099902199868,
      "compression_loss": 0.0,
      "distillation_loss": 0.024694666266441345,
      "epoch": 6.74,
      "learning_rate": 1.727723490081129e-07,
      "loss": 0.026,
      "step": 7099,
      "task_loss": 0.03818400576710701
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996108719297808,
      "compression_loss": 0.0,
      "distillation_loss": 0.04019855335354805,
      "epoch": 6.74,
      "learning_rate": 1.7152407453501219e-07,
      "loss": 0.0374,
      "step": 7100,
      "task_loss": 0.012513427063822746
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799611752309696,
      "compression_loss": 0.0,
      "distillation_loss": 0.028096789494156837,
      "epoch": 6.74,
      "learning_rate": 1.7028031027413737e-07,
      "loss": 0.0337,
      "step": 7101,
      "task_loss": 0.08449839055538177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799612631360736,
      "compression_loss": 0.0,
      "distillation_loss": 0.05807451158761978,
      "epoch": 6.74,
      "learning_rate": 1.6904105645142444e-07,
      "loss": 0.0659,
      "step": 7102,
      "task_loss": 0.13655823469161987
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996135090839047,
      "compression_loss": 0.0,
      "distillation_loss": 0.05028388276696205,
      "epoch": 6.75,
      "learning_rate": 1.6780631329199326e-07,
      "loss": 0.047,
      "step": 7103,
      "task_loss": 0.01732676476240158
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996143854802057,
      "compression_loss": 0.0,
      "distillation_loss": 0.01434339303523302,
      "epoch": 6.75,
      "learning_rate": 1.6657608102013667e-07,
      "loss": 0.0217,
      "step": 7104,
      "task_loss": 0.08795058727264404
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996152605506426,
      "compression_loss": 0.0,
      "distillation_loss": 0.026891184970736504,
      "epoch": 6.75,
      "learning_rate": 1.65350359859337e-07,
      "loss": 0.0254,
      "step": 7105,
      "task_loss": 0.011834867298603058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996161342962191,
      "compression_loss": 0.0,
      "distillation_loss": 0.058868955820798874,
      "epoch": 6.75,
      "learning_rate": 1.6412915003224672e-07,
      "loss": 0.056,
      "step": 7106,
      "task_loss": 0.030042661353945732
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799617006717939,
      "compression_loss": 0.0,
      "distillation_loss": 0.023804733529686928,
      "epoch": 6.75,
      "learning_rate": 1.629124517607078e-07,
      "loss": 0.022,
      "step": 7107,
      "task_loss": 0.006073350086808205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996178778168059,
      "compression_loss": 0.0,
      "distillation_loss": 0.025645965710282326,
      "epoch": 6.75,
      "learning_rate": 1.6170026526574066e-07,
      "loss": 0.0248,
      "step": 7108,
      "task_loss": 0.0175074003636837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996187475938235,
      "compression_loss": 0.0,
      "distillation_loss": 0.031395357102155685,
      "epoch": 6.75,
      "learning_rate": 1.6049259076753864e-07,
      "loss": 0.0353,
      "step": 7109,
      "task_loss": 0.07032643258571625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996196160499953,
      "compression_loss": 0.0,
      "distillation_loss": 0.025923658162355423,
      "epoch": 6.75,
      "learning_rate": 1.592894284854901e-07,
      "loss": 0.024,
      "step": 7110,
      "task_loss": 0.006603613495826721
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996204831863253,
      "compression_loss": 0.0,
      "distillation_loss": 0.11871407926082611,
      "epoch": 6.75,
      "learning_rate": 1.5809077863814803e-07,
      "loss": 0.117,
      "step": 7111,
      "task_loss": 0.10196692496538162
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799621349003817,
      "compression_loss": 0.0,
      "distillation_loss": 0.06971892714500427,
      "epoch": 6.75,
      "learning_rate": 1.5689664144325766e-07,
      "loss": 0.0639,
      "step": 7112,
      "task_loss": 0.011437831446528435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996222135034742,
      "compression_loss": 0.0,
      "distillation_loss": 0.0784948393702507,
      "epoch": 6.75,
      "learning_rate": 1.5570701711773717e-07,
      "loss": 0.0759,
      "step": 7113,
      "task_loss": 0.05259266868233681
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996230766863003,
      "compression_loss": 0.0,
      "distillation_loss": 0.03151058405637741,
      "epoch": 6.76,
      "learning_rate": 1.5452190587768867e-07,
      "loss": 0.0486,
      "step": 7114,
      "task_loss": 0.20285047590732574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996239385532993,
      "compression_loss": 0.0,
      "distillation_loss": 0.09722236543893814,
      "epoch": 6.76,
      "learning_rate": 1.5334130793839275e-07,
      "loss": 0.1164,
      "step": 7115,
      "task_loss": 0.2885318398475647
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996247991054748,
      "compression_loss": 0.0,
      "distillation_loss": 0.07263641804456711,
      "epoch": 6.76,
      "learning_rate": 1.5216522351431118e-07,
      "loss": 0.0823,
      "step": 7116,
      "task_loss": 0.16909703612327576
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996256583438304,
      "compression_loss": 0.0,
      "distillation_loss": 0.04487297311425209,
      "epoch": 6.76,
      "learning_rate": 1.5099365281908694e-07,
      "loss": 0.0531,
      "step": 7117,
      "task_loss": 0.12665140628814697
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996265162693698,
      "compression_loss": 0.0,
      "distillation_loss": 0.013267583213746548,
      "epoch": 6.76,
      "learning_rate": 1.4982659606553872e-07,
      "loss": 0.0206,
      "step": 7118,
      "task_loss": 0.08665129542350769
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996273728830966,
      "compression_loss": 0.0,
      "distillation_loss": 0.012607241049408913,
      "epoch": 6.76,
      "learning_rate": 1.4866405346566914e-07,
      "loss": 0.0118,
      "step": 7119,
      "task_loss": 0.004675716161727905
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996282281860148,
      "compression_loss": 0.0,
      "distillation_loss": 0.06657108664512634,
      "epoch": 6.76,
      "learning_rate": 1.4750602523065648e-07,
      "loss": 0.0707,
      "step": 7120,
      "task_loss": 0.10777979344129562
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996290821791276,
      "compression_loss": 0.0,
      "distillation_loss": 0.028429970145225525,
      "epoch": 6.76,
      "learning_rate": 1.463525115708686e-07,
      "loss": 0.0311,
      "step": 7121,
      "task_loss": 0.05511371046304703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996299348634391,
      "compression_loss": 0.0,
      "distillation_loss": 0.039049431681632996,
      "epoch": 6.76,
      "learning_rate": 1.4520351269584343e-07,
      "loss": 0.0477,
      "step": 7122,
      "task_loss": 0.12538807094097137
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996307862399529,
      "compression_loss": 0.0,
      "distillation_loss": 0.036447908729314804,
      "epoch": 6.76,
      "learning_rate": 1.4405902881430288e-07,
      "loss": 0.0352,
      "step": 7123,
      "task_loss": 0.023651236668229103
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996316363096726,
      "compression_loss": 0.0,
      "distillation_loss": 0.027506425976753235,
      "epoch": 6.77,
      "learning_rate": 1.4291906013414457e-07,
      "loss": 0.0269,
      "step": 7124,
      "task_loss": 0.02168971858918667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996324850736019,
      "compression_loss": 0.0,
      "distillation_loss": 0.02820620872080326,
      "epoch": 6.77,
      "learning_rate": 1.417836068624556e-07,
      "loss": 0.0268,
      "step": 7125,
      "task_loss": 0.014426644891500473
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996333325327444,
      "compression_loss": 0.0,
      "distillation_loss": 0.04337786138057709,
      "epoch": 6.77,
      "learning_rate": 1.406526692054877e-07,
      "loss": 0.0525,
      "step": 7126,
      "task_loss": 0.13505274057388306
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996341786881038,
      "compression_loss": 0.0,
      "distillation_loss": 0.031544819474220276,
      "epoch": 6.77,
      "learning_rate": 1.3952624736869036e-07,
      "loss": 0.0305,
      "step": 7127,
      "task_loss": 0.0206135306507349
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996350235406839,
      "compression_loss": 0.0,
      "distillation_loss": 0.027235832065343857,
      "epoch": 6.77,
      "learning_rate": 1.3840434155668047e-07,
      "loss": 0.0423,
      "step": 7128,
      "task_loss": 0.17793741822242737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996358670914885,
      "compression_loss": 0.0,
      "distillation_loss": 0.026448873803019524,
      "epoch": 6.77,
      "learning_rate": 1.3728695197325336e-07,
      "loss": 0.0328,
      "step": 7129,
      "task_loss": 0.08974019438028336
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996367093415209,
      "compression_loss": 0.0,
      "distillation_loss": 0.03855022042989731,
      "epoch": 6.77,
      "learning_rate": 1.361740788213911e-07,
      "loss": 0.0482,
      "step": 7130,
      "task_loss": 0.13530199229717255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996375502917851,
      "compression_loss": 0.0,
      "distillation_loss": 0.02457747608423233,
      "epoch": 6.77,
      "learning_rate": 1.3506572230325698e-07,
      "loss": 0.0226,
      "step": 7131,
      "task_loss": 0.004712225869297981
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996383899432846,
      "compression_loss": 0.0,
      "distillation_loss": 0.09410484880208969,
      "epoch": 6.77,
      "learning_rate": 1.3396188262018438e-07,
      "loss": 0.1019,
      "step": 7132,
      "task_loss": 0.17175397276878357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996392282970232,
      "compression_loss": 0.0,
      "distillation_loss": 0.05483167618513107,
      "epoch": 6.77,
      "learning_rate": 1.3286255997268793e-07,
      "loss": 0.0599,
      "step": 7133,
      "task_loss": 0.10567161440849304
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996400653540044,
      "compression_loss": 0.0,
      "distillation_loss": 0.03218763321638107,
      "epoch": 6.77,
      "learning_rate": 1.3176775456047175e-07,
      "loss": 0.0324,
      "step": 7134,
      "task_loss": 0.034330014139413834
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996409011152322,
      "compression_loss": 0.0,
      "distillation_loss": 0.038693610578775406,
      "epoch": 6.78,
      "learning_rate": 1.3067746658241008e-07,
      "loss": 0.0474,
      "step": 7135,
      "task_loss": 0.12591272592544556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79964173558171,
      "compression_loss": 0.0,
      "distillation_loss": 0.018780706450343132,
      "epoch": 6.78,
      "learning_rate": 1.2959169623655843e-07,
      "loss": 0.0281,
      "step": 7136,
      "task_loss": 0.11212365329265594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996425687544417,
      "compression_loss": 0.0,
      "distillation_loss": 0.021541312336921692,
      "epoch": 6.78,
      "learning_rate": 1.2851044372015064e-07,
      "loss": 0.0275,
      "step": 7137,
      "task_loss": 0.08120422065258026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996434006344307,
      "compression_loss": 0.0,
      "distillation_loss": 0.018733816221356392,
      "epoch": 6.78,
      "learning_rate": 1.2743370922960462e-07,
      "loss": 0.0175,
      "step": 7138,
      "task_loss": 0.006464965641498566
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799644231222681,
      "compression_loss": 0.0,
      "distillation_loss": 0.04339815676212311,
      "epoch": 6.78,
      "learning_rate": 1.263614929605139e-07,
      "loss": 0.0406,
      "step": 7139,
      "task_loss": 0.015392309054732323
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799645060520196,
      "compression_loss": 0.0,
      "distillation_loss": 0.01880636438727379,
      "epoch": 6.78,
      "learning_rate": 1.252937951076477e-07,
      "loss": 0.0175,
      "step": 7140,
      "task_loss": 0.005763756111264229
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996458885279796,
      "compression_loss": 0.0,
      "distillation_loss": 0.0439765602350235,
      "epoch": 6.78,
      "learning_rate": 1.2423061586496477e-07,
      "loss": 0.0415,
      "step": 7141,
      "task_loss": 0.019159667193889618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996467152470353,
      "compression_loss": 0.0,
      "distillation_loss": 0.036824434995651245,
      "epoch": 6.78,
      "learning_rate": 1.231719554255939e-07,
      "loss": 0.0418,
      "step": 7142,
      "task_loss": 0.08699777722358704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996475406783671,
      "compression_loss": 0.0,
      "distillation_loss": 0.014559917151927948,
      "epoch": 6.78,
      "learning_rate": 1.2211781398184242e-07,
      "loss": 0.0138,
      "step": 7143,
      "task_loss": 0.0064619556069374084
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996483648229782,
      "compression_loss": 0.0,
      "distillation_loss": 0.042142875492572784,
      "epoch": 6.78,
      "learning_rate": 1.2106819172520434e-07,
      "loss": 0.0426,
      "step": 7144,
      "task_loss": 0.046216074377298355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996491876818727,
      "compression_loss": 0.0,
      "distillation_loss": 0.023237306624650955,
      "epoch": 6.79,
      "learning_rate": 1.200230888463466e-07,
      "loss": 0.0304,
      "step": 7145,
      "task_loss": 0.0949011892080307
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799650009256054,
      "compression_loss": 0.0,
      "distillation_loss": 0.018296122550964355,
      "epoch": 6.79,
      "learning_rate": 1.1898250553512014e-07,
      "loss": 0.024,
      "step": 7146,
      "task_loss": 0.07504522800445557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799650829546526,
      "compression_loss": 0.0,
      "distillation_loss": 0.02411121129989624,
      "epoch": 6.79,
      "learning_rate": 1.1794644198054871e-07,
      "loss": 0.0222,
      "step": 7147,
      "task_loss": 0.0048682671040296555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996516485542922,
      "compression_loss": 0.0,
      "distillation_loss": 0.07223325222730637,
      "epoch": 6.79,
      "learning_rate": 1.1691489837083735e-07,
      "loss": 0.0702,
      "step": 7148,
      "task_loss": 0.052256107330322266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996524662803564,
      "compression_loss": 0.0,
      "distillation_loss": 0.0212749931961298,
      "epoch": 6.79,
      "learning_rate": 1.1588787489337505e-07,
      "loss": 0.0325,
      "step": 7149,
      "task_loss": 0.13337501883506775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996532827257222,
      "compression_loss": 0.0,
      "distillation_loss": 0.014818870462477207,
      "epoch": 6.79,
      "learning_rate": 1.1486537173472367e-07,
      "loss": 0.014,
      "step": 7150,
      "task_loss": 0.006277905777096748
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996540978913934,
      "compression_loss": 0.0,
      "distillation_loss": 0.05914817005395889,
      "epoch": 6.79,
      "learning_rate": 1.1384738908062631e-07,
      "loss": 0.0666,
      "step": 7151,
      "task_loss": 0.1340140402317047
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996549117783737,
      "compression_loss": 0.0,
      "distillation_loss": 0.044574763625860214,
      "epoch": 6.79,
      "learning_rate": 1.1283392711600171e-07,
      "loss": 0.0479,
      "step": 7152,
      "task_loss": 0.07772877812385559
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996557243876665,
      "compression_loss": 0.0,
      "distillation_loss": 0.0406227707862854,
      "epoch": 6.79,
      "learning_rate": 1.1182498602495539e-07,
      "loss": 0.0471,
      "step": 7153,
      "task_loss": 0.10571486502885818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996565357202758,
      "compression_loss": 0.0,
      "distillation_loss": 0.08192355930805206,
      "epoch": 6.79,
      "learning_rate": 1.1082056599076018e-07,
      "loss": 0.0774,
      "step": 7154,
      "task_loss": 0.03656609356403351
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996573457772052,
      "compression_loss": 0.0,
      "distillation_loss": 0.0265779010951519,
      "epoch": 6.79,
      "learning_rate": 1.0982066719587569e-07,
      "loss": 0.0401,
      "step": 7155,
      "task_loss": 0.16203996539115906
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996581545594582,
      "compression_loss": 0.0,
      "distillation_loss": 0.017745571210980415,
      "epoch": 6.8,
      "learning_rate": 1.0882528982194273e-07,
      "loss": 0.0164,
      "step": 7156,
      "task_loss": 0.0038829054683446884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996589620680388,
      "compression_loss": 0.0,
      "distillation_loss": 0.022481059655547142,
      "epoch": 6.8,
      "learning_rate": 1.0783443404976946e-07,
      "loss": 0.0208,
      "step": 7157,
      "task_loss": 0.005929671227931976
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996597683039504,
      "compression_loss": 0.0,
      "distillation_loss": 0.02317318134009838,
      "epoch": 6.8,
      "learning_rate": 1.0684810005935631e-07,
      "loss": 0.0214,
      "step": 7158,
      "task_loss": 0.005201876163482666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996605732681967,
      "compression_loss": 0.0,
      "distillation_loss": 0.05156789720058441,
      "epoch": 6.8,
      "learning_rate": 1.0586628802987108e-07,
      "loss": 0.0496,
      "step": 7159,
      "task_loss": 0.03225575387477875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996613769617816,
      "compression_loss": 0.0,
      "distillation_loss": 0.02196839265525341,
      "epoch": 6.8,
      "learning_rate": 1.0488899813966835e-07,
      "loss": 0.0267,
      "step": 7160,
      "task_loss": 0.06964989006519318
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996621793857086,
      "compression_loss": 0.0,
      "distillation_loss": 0.06468576192855835,
      "epoch": 6.8,
      "learning_rate": 1.0391623056627275e-07,
      "loss": 0.0656,
      "step": 7161,
      "task_loss": 0.07367479801177979
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996629805409815,
      "compression_loss": 0.0,
      "distillation_loss": 0.04920611530542374,
      "epoch": 6.8,
      "learning_rate": 1.0294798548639573e-07,
      "loss": 0.0528,
      "step": 7162,
      "task_loss": 0.08470079302787781
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996637804286039,
      "compression_loss": 0.0,
      "distillation_loss": 0.032296232879161835,
      "epoch": 6.8,
      "learning_rate": 1.0198426307592157e-07,
      "loss": 0.0307,
      "step": 7163,
      "task_loss": 0.01606859639286995
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996645790495794,
      "compression_loss": 0.0,
      "distillation_loss": 0.04291418194770813,
      "epoch": 6.8,
      "learning_rate": 1.010250635099158e-07,
      "loss": 0.0641,
      "step": 7164,
      "task_loss": 0.25430774688720703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996653764049119,
      "compression_loss": 0.0,
      "distillation_loss": 0.02546636387705803,
      "epoch": 6.8,
      "learning_rate": 1.0007038696262516e-07,
      "loss": 0.0302,
      "step": 7165,
      "task_loss": 0.07231911271810532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996661724956049,
      "compression_loss": 0.0,
      "distillation_loss": 0.028939777985215187,
      "epoch": 6.81,
      "learning_rate": 9.912023360746647e-08,
      "loss": 0.0268,
      "step": 7166,
      "task_loss": 0.007344096899032593
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996669673226621,
      "compression_loss": 0.0,
      "distillation_loss": 0.11444491147994995,
      "epoch": 6.81,
      "learning_rate": 9.817460361704056e-08,
      "loss": 0.1122,
      "step": 7167,
      "task_loss": 0.0920182541012764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996677608870872,
      "compression_loss": 0.0,
      "distillation_loss": 0.10061150789260864,
      "epoch": 6.81,
      "learning_rate": 9.723349716312668e-08,
      "loss": 0.0963,
      "step": 7168,
      "task_loss": 0.05738703906536102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996685531898841,
      "compression_loss": 0.0,
      "distillation_loss": 0.08503048866987228,
      "epoch": 6.81,
      "learning_rate": 9.629691441667976e-08,
      "loss": 0.093,
      "step": 7169,
      "task_loss": 0.1647735834121704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996693442320562,
      "compression_loss": 0.0,
      "distillation_loss": 0.028403718024492264,
      "epoch": 6.81,
      "learning_rate": 9.53648555478387e-08,
      "loss": 0.0262,
      "step": 7170,
      "task_loss": 0.006750069558620453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996701340146072,
      "compression_loss": 0.0,
      "distillation_loss": 0.01994595304131508,
      "epoch": 6.81,
      "learning_rate": 9.443732072591249e-08,
      "loss": 0.0188,
      "step": 7171,
      "task_loss": 0.008648447692394257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996709225385409,
      "compression_loss": 0.0,
      "distillation_loss": 0.06424511969089508,
      "epoch": 6.81,
      "learning_rate": 9.351431011939138e-08,
      "loss": 0.0761,
      "step": 7172,
      "task_loss": 0.18302549421787262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799671709804861,
      "compression_loss": 0.0,
      "distillation_loss": 0.020225565880537033,
      "epoch": 6.81,
      "learning_rate": 9.259582389594956e-08,
      "loss": 0.0185,
      "step": 7173,
      "task_loss": 0.002962857484817505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799672495814571,
      "compression_loss": 0.0,
      "distillation_loss": 0.07207255065441132,
      "epoch": 6.81,
      "learning_rate": 9.168186222243136e-08,
      "loss": 0.0706,
      "step": 7174,
      "task_loss": 0.05724616348743439
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996732805686748,
      "compression_loss": 0.0,
      "distillation_loss": 0.02577575296163559,
      "epoch": 6.81,
      "learning_rate": 9.077242526485952e-08,
      "loss": 0.027,
      "step": 7175,
      "task_loss": 0.03761523589491844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799674064068176,
      "compression_loss": 0.0,
      "distillation_loss": 0.03172352910041809,
      "epoch": 6.81,
      "learning_rate": 8.986751318844355e-08,
      "loss": 0.0399,
      "step": 7176,
      "task_loss": 0.11340309679508209
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996748463140781,
      "compression_loss": 0.0,
      "distillation_loss": 0.03157733753323555,
      "epoch": 6.82,
      "learning_rate": 8.896712615756308e-08,
      "loss": 0.0311,
      "step": 7177,
      "task_loss": 0.02710825763642788
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996756273073852,
      "compression_loss": 0.0,
      "distillation_loss": 0.01715652272105217,
      "epoch": 6.82,
      "learning_rate": 8.807126433577617e-08,
      "loss": 0.0165,
      "step": 7178,
      "task_loss": 0.010277284309267998
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996764070491006,
      "compression_loss": 0.0,
      "distillation_loss": 0.03965865075588226,
      "epoch": 6.82,
      "learning_rate": 8.717992788582207e-08,
      "loss": 0.0431,
      "step": 7179,
      "task_loss": 0.07392223924398422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799677185540228,
      "compression_loss": 0.0,
      "distillation_loss": 0.04710112512111664,
      "epoch": 6.82,
      "learning_rate": 8.629311696961295e-08,
      "loss": 0.0486,
      "step": 7180,
      "task_loss": 0.06256051361560822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996779627817714,
      "compression_loss": 0.0,
      "distillation_loss": 0.04641801118850708,
      "epoch": 6.82,
      "learning_rate": 8.541083174824771e-08,
      "loss": 0.0597,
      "step": 7181,
      "task_loss": 0.17936520278453827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996787387747342,
      "compression_loss": 0.0,
      "distillation_loss": 0.017239782959222794,
      "epoch": 6.82,
      "learning_rate": 8.453307238199259e-08,
      "loss": 0.0159,
      "step": 7182,
      "task_loss": 0.003765510395169258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996795135201201,
      "compression_loss": 0.0,
      "distillation_loss": 0.13984356820583344,
      "epoch": 6.82,
      "learning_rate": 8.365983903030061e-08,
      "loss": 0.139,
      "step": 7183,
      "task_loss": 0.13107913732528687
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996802870189329,
      "compression_loss": 0.0,
      "distillation_loss": 0.07437920570373535,
      "epoch": 6.82,
      "learning_rate": 8.279113185179488e-08,
      "loss": 0.0826,
      "step": 7184,
      "task_loss": 0.15642966330051422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996810592721761,
      "compression_loss": 0.0,
      "distillation_loss": 0.036259036511182785,
      "epoch": 6.82,
      "learning_rate": 8.192695100428527e-08,
      "loss": 0.0332,
      "step": 7185,
      "task_loss": 0.005903052166104317
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996818302808537,
      "compression_loss": 0.0,
      "distillation_loss": 0.018497422337532043,
      "epoch": 6.82,
      "learning_rate": 8.106729664475176e-08,
      "loss": 0.0241,
      "step": 7186,
      "task_loss": 0.07415502518415451
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799682600045969,
      "compression_loss": 0.0,
      "distillation_loss": 0.03047127276659012,
      "epoch": 6.83,
      "learning_rate": 8.021216892935279e-08,
      "loss": 0.0313,
      "step": 7187,
      "task_loss": 0.038899097591638565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799683368568526,
      "compression_loss": 0.0,
      "distillation_loss": 0.053710635751485825,
      "epoch": 6.83,
      "learning_rate": 7.936156801342797e-08,
      "loss": 0.0536,
      "step": 7188,
      "task_loss": 0.052580974996089935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996841358495282,
      "compression_loss": 0.0,
      "distillation_loss": 0.06118376553058624,
      "epoch": 6.83,
      "learning_rate": 7.851549405149539e-08,
      "loss": 0.0601,
      "step": 7189,
      "task_loss": 0.050785936415195465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996849018899793,
      "compression_loss": 0.0,
      "distillation_loss": 0.02713852934539318,
      "epoch": 6.83,
      "learning_rate": 7.767394719724597e-08,
      "loss": 0.0425,
      "step": 7190,
      "task_loss": 0.18106593191623688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996856666908831,
      "compression_loss": 0.0,
      "distillation_loss": 0.06980240345001221,
      "epoch": 6.83,
      "learning_rate": 7.683692760355187e-08,
      "loss": 0.076,
      "step": 7191,
      "task_loss": 0.13159391283988953
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996864302532432,
      "compression_loss": 0.0,
      "distillation_loss": 0.019683901220560074,
      "epoch": 6.83,
      "learning_rate": 7.600443542246094e-08,
      "loss": 0.0184,
      "step": 7192,
      "task_loss": 0.00683322548866272
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996871925780632,
      "compression_loss": 0.0,
      "distillation_loss": 0.02726883627474308,
      "epoch": 6.83,
      "learning_rate": 7.51764708051994e-08,
      "loss": 0.0323,
      "step": 7193,
      "task_loss": 0.0772031769156456
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996879536663469,
      "compression_loss": 0.0,
      "distillation_loss": 0.15833452343940735,
      "epoch": 6.83,
      "learning_rate": 7.435303390216919e-08,
      "loss": 0.1523,
      "step": 7194,
      "task_loss": 0.09784331917762756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799688713519098,
      "compression_loss": 0.0,
      "distillation_loss": 0.02268981747329235,
      "epoch": 6.83,
      "learning_rate": 7.353412486295619e-08,
      "loss": 0.0208,
      "step": 7195,
      "task_loss": 0.004109309986233711
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79968947213732,
      "compression_loss": 0.0,
      "distillation_loss": 0.024118095636367798,
      "epoch": 6.83,
      "learning_rate": 7.271974383631642e-08,
      "loss": 0.0263,
      "step": 7196,
      "task_loss": 0.04595687985420227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996902295220168,
      "compression_loss": 0.0,
      "distillation_loss": 0.028797656297683716,
      "epoch": 6.83,
      "learning_rate": 7.19098909701843e-08,
      "loss": 0.0415,
      "step": 7197,
      "task_loss": 0.1560361236333847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799690985674192,
      "compression_loss": 0.0,
      "distillation_loss": 0.03179541975259781,
      "epoch": 6.84,
      "learning_rate": 7.110456641167829e-08,
      "loss": 0.0305,
      "step": 7198,
      "task_loss": 0.018481941893696785
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996917405948492,
      "compression_loss": 0.0,
      "distillation_loss": 0.01870710775256157,
      "epoch": 6.84,
      "learning_rate": 7.030377030708412e-08,
      "loss": 0.0181,
      "step": 7199,
      "task_loss": 0.012378891929984093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996924942849922,
      "compression_loss": 0.0,
      "distillation_loss": 0.06044170260429382,
      "epoch": 6.84,
      "learning_rate": 6.950750280187435e-08,
      "loss": 0.0777,
      "step": 7200,
      "task_loss": 0.23314093053340912
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996932467456246,
      "compression_loss": 0.0,
      "distillation_loss": 0.05844536051154137,
      "epoch": 6.84,
      "learning_rate": 6.871576404069158e-08,
      "loss": 0.0756,
      "step": 7201,
      "task_loss": 0.23015083372592926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996939979777502,
      "compression_loss": 0.0,
      "distillation_loss": 0.01944398134946823,
      "epoch": 6.84,
      "learning_rate": 6.792855416736243e-08,
      "loss": 0.0264,
      "step": 7202,
      "task_loss": 0.08910975605249405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996947479823725,
      "compression_loss": 0.0,
      "distillation_loss": 0.024285856634378433,
      "epoch": 6.84,
      "learning_rate": 6.714587332488364e-08,
      "loss": 0.0317,
      "step": 7203,
      "task_loss": 0.09848549962043762
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996954967604953,
      "compression_loss": 0.0,
      "distillation_loss": 0.06020985543727875,
      "epoch": 6.84,
      "learning_rate": 6.63677216554387e-08,
      "loss": 0.0604,
      "step": 7204,
      "task_loss": 0.062497399747371674
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996962443131224,
      "compression_loss": 0.0,
      "distillation_loss": 0.03330262005329132,
      "epoch": 6.84,
      "learning_rate": 6.559409930037563e-08,
      "loss": 0.0311,
      "step": 7205,
      "task_loss": 0.011487048119306564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996969906412571,
      "compression_loss": 0.0,
      "distillation_loss": 0.08118142187595367,
      "epoch": 6.84,
      "learning_rate": 6.482500640022926e-08,
      "loss": 0.0848,
      "step": 7206,
      "task_loss": 0.11695171892642975
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996977357459035,
      "compression_loss": 0.0,
      "distillation_loss": 0.030360179021954536,
      "epoch": 6.84,
      "learning_rate": 6.406044309471005e-08,
      "loss": 0.0329,
      "step": 7207,
      "task_loss": 0.05534449219703674
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996984796280651,
      "compression_loss": 0.0,
      "distillation_loss": 0.08223249763250351,
      "epoch": 6.85,
      "learning_rate": 6.406044309471005e-08,
      "loss": 0.0893,
      "step": 7208,
      "task_loss": 0.15278904139995575
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996992222887456,
      "compression_loss": 0.0,
      "distillation_loss": 0.022313468158245087,
      "epoch": 6.85,
      "learning_rate": 6.330040952270688e-08,
      "loss": 0.0244,
      "step": 7209,
      "task_loss": 0.04346153512597084
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7996999637289486,
      "compression_loss": 0.0,
      "distillation_loss": 0.0184345506131649,
      "epoch": 6.85,
      "learning_rate": 6.254490582227879e-08,
      "loss": 0.017,
      "step": 7210,
      "task_loss": 0.004323702305555344
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799700703949678,
      "compression_loss": 0.0,
      "distillation_loss": 0.020290520042181015,
      "epoch": 6.85,
      "learning_rate": 6.179393213066875e-08,
      "loss": 0.0271,
      "step": 7211,
      "task_loss": 0.08879521489143372
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997014429519371,
      "compression_loss": 0.0,
      "distillation_loss": 0.020489336922764778,
      "epoch": 6.85,
      "learning_rate": 6.104748858429266e-08,
      "loss": 0.0298,
      "step": 7212,
      "task_loss": 0.11318296194076538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79970218073673,
      "compression_loss": 0.0,
      "distillation_loss": 0.05874666944146156,
      "epoch": 6.85,
      "learning_rate": 6.030557531875036e-08,
      "loss": 0.0728,
      "step": 7213,
      "task_loss": 0.1989019811153412
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997029173050602,
      "compression_loss": 0.0,
      "distillation_loss": 0.03400994837284088,
      "epoch": 6.85,
      "learning_rate": 5.956819246881185e-08,
      "loss": 0.0417,
      "step": 7214,
      "task_loss": 0.11090053617954254
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997036526579313,
      "compression_loss": 0.0,
      "distillation_loss": 0.02666945569217205,
      "epoch": 6.85,
      "learning_rate": 5.8835340168422734e-08,
      "loss": 0.027,
      "step": 7215,
      "task_loss": 0.029919061809778214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997043867963471,
      "compression_loss": 0.0,
      "distillation_loss": 0.032640133053064346,
      "epoch": 6.85,
      "learning_rate": 5.8107018550712656e-08,
      "loss": 0.0301,
      "step": 7216,
      "task_loss": 0.0070366039872169495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997051197213113,
      "compression_loss": 0.0,
      "distillation_loss": 0.02398163080215454,
      "epoch": 6.85,
      "learning_rate": 5.7383227747984105e-08,
      "loss": 0.0315,
      "step": 7217,
      "task_loss": 0.09866875410079956
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997058514338276,
      "compression_loss": 0.0,
      "distillation_loss": 0.01552260760217905,
      "epoch": 6.85,
      "learning_rate": 5.6663967891718015e-08,
      "loss": 0.0146,
      "step": 7218,
      "task_loss": 0.006190884858369827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997065819348994,
      "compression_loss": 0.0,
      "distillation_loss": 0.06299366801977158,
      "epoch": 6.86,
      "learning_rate": 5.5949239112570997e-08,
      "loss": 0.0748,
      "step": 7219,
      "task_loss": 0.18070363998413086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997073112255307,
      "compression_loss": 0.0,
      "distillation_loss": 0.056522756814956665,
      "epoch": 6.86,
      "learning_rate": 5.523904154037529e-08,
      "loss": 0.0572,
      "step": 7220,
      "task_loss": 0.06307417154312134
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997080393067251,
      "compression_loss": 0.0,
      "distillation_loss": 0.02577255666255951,
      "epoch": 6.86,
      "learning_rate": 5.453337530414437e-08,
      "loss": 0.0269,
      "step": 7221,
      "task_loss": 0.03711457550525665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997087661794863,
      "compression_loss": 0.0,
      "distillation_loss": 0.024436548352241516,
      "epoch": 6.86,
      "learning_rate": 5.383224053206459e-08,
      "loss": 0.0283,
      "step": 7222,
      "task_loss": 0.06333892792463303
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997094918448179,
      "compression_loss": 0.0,
      "distillation_loss": 0.040110617876052856,
      "epoch": 6.86,
      "learning_rate": 5.313563735149796e-08,
      "loss": 0.0419,
      "step": 7223,
      "task_loss": 0.057536765933036804
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997102163037235,
      "compression_loss": 0.0,
      "distillation_loss": 0.026415549218654633,
      "epoch": 6.86,
      "learning_rate": 5.2443565888990466e-08,
      "loss": 0.0339,
      "step": 7224,
      "task_loss": 0.10137321054935455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997109395572071,
      "compression_loss": 0.0,
      "distillation_loss": 0.05468286946415901,
      "epoch": 6.86,
      "learning_rate": 5.1756026270258215e-08,
      "loss": 0.0534,
      "step": 7225,
      "task_loss": 0.0421706885099411
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799711661606272,
      "compression_loss": 0.0,
      "distillation_loss": 0.04075568914413452,
      "epoch": 6.86,
      "learning_rate": 5.107301862019575e-08,
      "loss": 0.0453,
      "step": 7226,
      "task_loss": 0.08584432303905487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997123824519222,
      "compression_loss": 0.0,
      "distillation_loss": 0.04388091713190079,
      "epoch": 6.86,
      "learning_rate": 5.0394543062873276e-08,
      "loss": 0.053,
      "step": 7227,
      "task_loss": 0.13531340658664703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997131020951613,
      "compression_loss": 0.0,
      "distillation_loss": 0.03439214825630188,
      "epoch": 6.86,
      "learning_rate": 4.972059972154219e-08,
      "loss": 0.0332,
      "step": 7228,
      "task_loss": 0.021979160606861115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997138205369929,
      "compression_loss": 0.0,
      "distillation_loss": 0.034759216010570526,
      "epoch": 6.87,
      "learning_rate": 4.905118871862402e-08,
      "loss": 0.038,
      "step": 7229,
      "task_loss": 0.06708045303821564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997145377784206,
      "compression_loss": 0.0,
      "distillation_loss": 0.08172931522130966,
      "epoch": 6.87,
      "learning_rate": 4.838631017572426e-08,
      "loss": 0.0904,
      "step": 7230,
      "task_loss": 0.16795513033866882
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997152538204483,
      "compression_loss": 0.0,
      "distillation_loss": 0.027006901800632477,
      "epoch": 6.87,
      "learning_rate": 4.772596421361852e-08,
      "loss": 0.0248,
      "step": 7231,
      "task_loss": 0.0044814664870500565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997159686640796,
      "compression_loss": 0.0,
      "distillation_loss": 0.053238674998283386,
      "epoch": 6.87,
      "learning_rate": 4.7070150952263634e-08,
      "loss": 0.068,
      "step": 7232,
      "task_loss": 0.2006755918264389
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997166823103181,
      "compression_loss": 0.0,
      "distillation_loss": 0.04943888634443283,
      "epoch": 6.87,
      "learning_rate": 4.64188705107893e-08,
      "loss": 0.0456,
      "step": 7233,
      "task_loss": 0.011105941608548164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997173947601676,
      "compression_loss": 0.0,
      "distillation_loss": 0.025495588779449463,
      "epoch": 6.87,
      "learning_rate": 4.577212300750644e-08,
      "loss": 0.0365,
      "step": 7234,
      "task_loss": 0.1354934424161911
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997181060146317,
      "compression_loss": 0.0,
      "distillation_loss": 0.021435732021927834,
      "epoch": 6.87,
      "learning_rate": 4.5129908559896075e-08,
      "loss": 0.0301,
      "step": 7235,
      "task_loss": 0.10771320015192032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997188160747141,
      "compression_loss": 0.0,
      "distillation_loss": 0.03189963847398758,
      "epoch": 6.87,
      "learning_rate": 4.449222728462599e-08,
      "loss": 0.0375,
      "step": 7236,
      "task_loss": 0.08780453354120255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997195249414185,
      "compression_loss": 0.0,
      "distillation_loss": 0.08029976487159729,
      "epoch": 6.87,
      "learning_rate": 4.3859079297525754e-08,
      "loss": 0.0884,
      "step": 7237,
      "task_loss": 0.1608220934867859
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997202326157485,
      "compression_loss": 0.0,
      "distillation_loss": 0.07394590228796005,
      "epoch": 6.87,
      "learning_rate": 4.323046471361447e-08,
      "loss": 0.0696,
      "step": 7238,
      "task_loss": 0.0303100124001503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799720939098708,
      "compression_loss": 0.0,
      "distillation_loss": 0.08067163825035095,
      "epoch": 6.87,
      "learning_rate": 4.2606383647084134e-08,
      "loss": 0.0771,
      "step": 7239,
      "task_loss": 0.04478314146399498
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997216443913004,
      "compression_loss": 0.0,
      "distillation_loss": 0.0285223089158535,
      "epoch": 6.88,
      "learning_rate": 4.198683621129962e-08,
      "loss": 0.0336,
      "step": 7240,
      "task_loss": 0.07915632426738739
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997223484945296,
      "compression_loss": 0.0,
      "distillation_loss": 0.02331477962434292,
      "epoch": 6.88,
      "learning_rate": 4.1371822518804224e-08,
      "loss": 0.0306,
      "step": 7241,
      "task_loss": 0.09588571637868881
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997230514093993,
      "compression_loss": 0.0,
      "distillation_loss": 0.07636144757270813,
      "epoch": 6.88,
      "learning_rate": 4.0761342681319706e-08,
      "loss": 0.0866,
      "step": 7242,
      "task_loss": 0.1786966323852539
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997237531369129,
      "compression_loss": 0.0,
      "distillation_loss": 0.08760742098093033,
      "epoch": 6.88,
      "learning_rate": 4.015539680974345e-08,
      "loss": 0.0921,
      "step": 7243,
      "task_loss": 0.13226494193077087
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997244536780743,
      "compression_loss": 0.0,
      "distillation_loss": 0.07857417315244675,
      "epoch": 6.88,
      "learning_rate": 3.955398501414576e-08,
      "loss": 0.0795,
      "step": 7244,
      "task_loss": 0.08820837736129761
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997251530338872,
      "compression_loss": 0.0,
      "distillation_loss": 0.024933654814958572,
      "epoch": 6.88,
      "learning_rate": 3.895710740378089e-08,
      "loss": 0.0264,
      "step": 7245,
      "task_loss": 0.03946792706847191
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997258512053552,
      "compression_loss": 0.0,
      "distillation_loss": 0.0409945584833622,
      "epoch": 6.88,
      "learning_rate": 3.8364764087067685e-08,
      "loss": 0.041,
      "step": 7246,
      "task_loss": 0.04090442508459091
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799726548193482,
      "compression_loss": 0.0,
      "distillation_loss": 0.1333804726600647,
      "epoch": 6.88,
      "learning_rate": 3.777695517161173e-08,
      "loss": 0.1276,
      "step": 7247,
      "task_loss": 0.07556381076574326
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997272439992713,
      "compression_loss": 0.0,
      "distillation_loss": 0.03527265042066574,
      "epoch": 6.88,
      "learning_rate": 3.7193680764191496e-08,
      "loss": 0.0327,
      "step": 7248,
      "task_loss": 0.00909445621073246
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997279386237268,
      "compression_loss": 0.0,
      "distillation_loss": 0.0798153355717659,
      "epoch": 6.88,
      "learning_rate": 3.66149409707639e-08,
      "loss": 0.0749,
      "step": 7249,
      "task_loss": 0.031093858182430267
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997286320678522,
      "compression_loss": 0.0,
      "distillation_loss": 0.04000500962138176,
      "epoch": 6.89,
      "learning_rate": 3.604073589645596e-08,
      "loss": 0.0366,
      "step": 7250,
      "task_loss": 0.00630572997033596
    },
    {
      "epoch": 6.89,
      "eval_accuracy": 0.8956422018348624,
      "eval_loss": 0.4339929223060608,
      "eval_runtime": 18.2364,
      "eval_samples_per_second": 47.816,
      "eval_steps_per_second": 5.977,
      "step": 7250
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997293243326511,
      "compression_loss": 0.0,
      "distillation_loss": 0.024513155221939087,
      "epoch": 6.89,
      "learning_rate": 3.547106564557312e-08,
      "loss": 0.0265,
      "step": 7251,
      "task_loss": 0.0446399450302124
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997300154191272,
      "compression_loss": 0.0,
      "distillation_loss": 0.02782837674021721,
      "epoch": 6.89,
      "learning_rate": 3.490593032160483e-08,
      "loss": 0.0319,
      "step": 7252,
      "task_loss": 0.06867832690477371
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997307053282843,
      "compression_loss": 0.0,
      "distillation_loss": 0.02256028726696968,
      "epoch": 6.89,
      "learning_rate": 3.4345330027207856e-08,
      "loss": 0.0274,
      "step": 7253,
      "task_loss": 0.07064933329820633
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997313940611258,
      "compression_loss": 0.0,
      "distillation_loss": 0.047051578760147095,
      "epoch": 6.89,
      "learning_rate": 3.378926486421463e-08,
      "loss": 0.061,
      "step": 7254,
      "task_loss": 0.18665584921836853
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997320816186557,
      "compression_loss": 0.0,
      "distillation_loss": 0.016445623710751534,
      "epoch": 6.89,
      "learning_rate": 3.3237734933641574e-08,
      "loss": 0.0218,
      "step": 7255,
      "task_loss": 0.07008873671293259
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997327680018775,
      "compression_loss": 0.0,
      "distillation_loss": 0.03820062056183815,
      "epoch": 6.89,
      "learning_rate": 3.269074033567798e-08,
      "loss": 0.0358,
      "step": 7256,
      "task_loss": 0.014143818989396095
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799733453211795,
      "compression_loss": 0.0,
      "distillation_loss": 0.03633254021406174,
      "epoch": 6.89,
      "learning_rate": 3.21482811696805e-08,
      "loss": 0.0364,
      "step": 7257,
      "task_loss": 0.036846742033958435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997341372494118,
      "compression_loss": 0.0,
      "distillation_loss": 0.03856148198246956,
      "epoch": 6.89,
      "learning_rate": 3.161035753419805e-08,
      "loss": 0.0463,
      "step": 7258,
      "task_loss": 0.11632915586233139
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997348201157316,
      "compression_loss": 0.0,
      "distillation_loss": 0.018683161586523056,
      "epoch": 6.89,
      "learning_rate": 3.107696952694139e-08,
      "loss": 0.0372,
      "step": 7259,
      "task_loss": 0.20371021330356598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799735501811758,
      "compression_loss": 0.0,
      "distillation_loss": 0.018803803250193596,
      "epoch": 6.89,
      "learning_rate": 3.054811724480522e-08,
      "loss": 0.0181,
      "step": 7260,
      "task_loss": 0.0114500243216753
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997361823384949,
      "compression_loss": 0.0,
      "distillation_loss": 0.014816414564847946,
      "epoch": 6.9,
      "learning_rate": 3.0023800783857134e-08,
      "loss": 0.0141,
      "step": 7261,
      "task_loss": 0.0073722414672374725
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997368616969457,
      "compression_loss": 0.0,
      "distillation_loss": 0.020906083285808563,
      "epoch": 6.9,
      "learning_rate": 2.950402023934318e-08,
      "loss": 0.0235,
      "step": 7262,
      "task_loss": 0.046787478029727936
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997375398881144,
      "compression_loss": 0.0,
      "distillation_loss": 0.045058127492666245,
      "epoch": 6.9,
      "learning_rate": 2.898877570568226e-08,
      "loss": 0.0595,
      "step": 7263,
      "task_loss": 0.1898675560951233
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997382169130043,
      "compression_loss": 0.0,
      "distillation_loss": 0.050636500120162964,
      "epoch": 6.9,
      "learning_rate": 2.8478067276471733e-08,
      "loss": 0.0479,
      "step": 7264,
      "task_loss": 0.02349741943180561
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997388927726194,
      "compression_loss": 0.0,
      "distillation_loss": 0.049480557441711426,
      "epoch": 6.9,
      "learning_rate": 2.7971895044487385e-08,
      "loss": 0.0677,
      "step": 7265,
      "task_loss": 0.23184102773666382
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997395674679633,
      "compression_loss": 0.0,
      "distillation_loss": 0.10401052981615067,
      "epoch": 6.9,
      "learning_rate": 2.7470259101672336e-08,
      "loss": 0.0976,
      "step": 7266,
      "task_loss": 0.04024471715092659
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997402410000396,
      "compression_loss": 0.0,
      "distillation_loss": 0.018137693405151367,
      "epoch": 6.9,
      "learning_rate": 2.6973159539153693e-08,
      "loss": 0.0169,
      "step": 7267,
      "task_loss": 0.006021425127983093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799740913369852,
      "compression_loss": 0.0,
      "distillation_loss": 0.024599701166152954,
      "epoch": 6.9,
      "learning_rate": 2.648059644723144e-08,
      "loss": 0.0234,
      "step": 7268,
      "task_loss": 0.012996546924114227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997415845784043,
      "compression_loss": 0.0,
      "distillation_loss": 0.030850766226649284,
      "epoch": 6.9,
      "learning_rate": 2.5992569915384014e-08,
      "loss": 0.051,
      "step": 7269,
      "task_loss": 0.2318606823682785
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997422546267,
      "compression_loss": 0.0,
      "distillation_loss": 0.16657190024852753,
      "epoch": 6.9,
      "learning_rate": 2.550908003226271e-08,
      "loss": 0.1601,
      "step": 7270,
      "task_loss": 0.10178233683109283
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997429235157429,
      "compression_loss": 0.0,
      "distillation_loss": 0.031046094372868538,
      "epoch": 6.91,
      "learning_rate": 2.5030126885694506e-08,
      "loss": 0.0305,
      "step": 7271,
      "task_loss": 0.02560516819357872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997435912465368,
      "compression_loss": 0.0,
      "distillation_loss": 0.018669994547963142,
      "epoch": 6.91,
      "learning_rate": 2.4555710562684796e-08,
      "loss": 0.0276,
      "step": 7272,
      "task_loss": 0.10760626196861267
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997442578200852,
      "compression_loss": 0.0,
      "distillation_loss": 0.04650671035051346,
      "epoch": 6.91,
      "learning_rate": 2.408583114941465e-08,
      "loss": 0.0566,
      "step": 7273,
      "task_loss": 0.14748124778270721
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997449232373918,
      "compression_loss": 0.0,
      "distillation_loss": 0.0697929635643959,
      "epoch": 6.91,
      "learning_rate": 2.3620488731235234e-08,
      "loss": 0.0814,
      "step": 7274,
      "task_loss": 0.18626686930656433
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997455874994602,
      "compression_loss": 0.0,
      "distillation_loss": 0.03916598856449127,
      "epoch": 6.91,
      "learning_rate": 2.3159683392684483e-08,
      "loss": 0.0364,
      "step": 7275,
      "task_loss": 0.011343579739332199
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997462506072943,
      "compression_loss": 0.0,
      "distillation_loss": 0.05003172531723976,
      "epoch": 6.91,
      "learning_rate": 2.2703415217464885e-08,
      "loss": 0.0553,
      "step": 7276,
      "task_loss": 0.10285012423992157
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997469125618977,
      "compression_loss": 0.0,
      "distillation_loss": 0.05633166432380676,
      "epoch": 6.91,
      "learning_rate": 2.2251684288462915e-08,
      "loss": 0.0563,
      "step": 7277,
      "task_loss": 0.055823661386966705
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799747573364274,
      "compression_loss": 0.0,
      "distillation_loss": 0.02220912277698517,
      "epoch": 6.91,
      "learning_rate": 2.180449068773793e-08,
      "loss": 0.0213,
      "step": 7278,
      "task_loss": 0.013472869992256165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799748233015427,
      "compression_loss": 0.0,
      "distillation_loss": 0.03192361444234848,
      "epoch": 6.91,
      "learning_rate": 2.136183449652218e-08,
      "loss": 0.0313,
      "step": 7279,
      "task_loss": 0.025719383731484413
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997488915163603,
      "compression_loss": 0.0,
      "distillation_loss": 0.08539696782827377,
      "epoch": 6.91,
      "learning_rate": 2.0923715795229115e-08,
      "loss": 0.0879,
      "step": 7280,
      "task_loss": 0.11043383926153183
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997495488680776,
      "compression_loss": 0.0,
      "distillation_loss": 0.050244055688381195,
      "epoch": 6.91,
      "learning_rate": 2.0490134663442295e-08,
      "loss": 0.0595,
      "step": 7281,
      "task_loss": 0.14261166751384735
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997502050715826,
      "compression_loss": 0.0,
      "distillation_loss": 0.020430155098438263,
      "epoch": 6.92,
      "learning_rate": 2.006109117992372e-08,
      "loss": 0.0187,
      "step": 7282,
      "task_loss": 0.003576911985874176
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997508601278789,
      "compression_loss": 0.0,
      "distillation_loss": 0.07387155294418335,
      "epoch": 6.92,
      "learning_rate": 1.9636585422616593e-08,
      "loss": 0.0831,
      "step": 7283,
      "task_loss": 0.16649088263511658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997515140379703,
      "compression_loss": 0.0,
      "distillation_loss": 0.024126257747411728,
      "epoch": 6.92,
      "learning_rate": 1.9216617468625908e-08,
      "loss": 0.0239,
      "step": 7284,
      "task_loss": 0.022267458960413933
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997521668028604,
      "compression_loss": 0.0,
      "distillation_loss": 0.1891712099313736,
      "epoch": 6.92,
      "learning_rate": 1.8801187394248965e-08,
      "loss": 0.1828,
      "step": 7285,
      "task_loss": 0.12499140202999115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799752818423553,
      "compression_loss": 0.0,
      "distillation_loss": 0.02490517869591713,
      "epoch": 6.92,
      "learning_rate": 1.8390295274944846e-08,
      "loss": 0.0272,
      "step": 7286,
      "task_loss": 0.048274118453264236
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997534689010516,
      "compression_loss": 0.0,
      "distillation_loss": 0.031054774299263954,
      "epoch": 6.92,
      "learning_rate": 1.7983941185356622e-08,
      "loss": 0.029,
      "step": 7287,
      "task_loss": 0.010997863486409187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79975411823636,
      "compression_loss": 0.0,
      "distillation_loss": 0.02562297135591507,
      "epoch": 6.92,
      "learning_rate": 1.7582125199303023e-08,
      "loss": 0.0392,
      "step": 7288,
      "task_loss": 0.16150373220443726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997547664304819,
      "compression_loss": 0.0,
      "distillation_loss": 0.06376007199287415,
      "epoch": 6.92,
      "learning_rate": 1.7184847389770108e-08,
      "loss": 0.0598,
      "step": 7289,
      "task_loss": 0.02367803268134594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997554134844209,
      "compression_loss": 0.0,
      "distillation_loss": 0.014210294932126999,
      "epoch": 6.92,
      "learning_rate": 1.679210782892793e-08,
      "loss": 0.0297,
      "step": 7290,
      "task_loss": 0.1686396598815918
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997560593991807,
      "compression_loss": 0.0,
      "distillation_loss": 0.02178233675658703,
      "epoch": 6.92,
      "learning_rate": 1.6403906588122185e-08,
      "loss": 0.0201,
      "step": 7291,
      "task_loss": 0.005438664928078651
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799756704175765,
      "compression_loss": 0.0,
      "distillation_loss": 0.055342864245176315,
      "epoch": 6.92,
      "learning_rate": 1.6020243737865927e-08,
      "loss": 0.0638,
      "step": 7292,
      "task_loss": 0.1397797167301178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997573478151776,
      "compression_loss": 0.0,
      "distillation_loss": 0.025284867733716965,
      "epoch": 6.93,
      "learning_rate": 1.5641119347858947e-08,
      "loss": 0.024,
      "step": 7293,
      "task_loss": 0.012278718873858452
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799757990318422,
      "compression_loss": 0.0,
      "distillation_loss": 0.013603068888187408,
      "epoch": 6.93,
      "learning_rate": 1.526653348696838e-08,
      "loss": 0.0152,
      "step": 7294,
      "task_loss": 0.029902005568146706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799758631686502,
      "compression_loss": 0.0,
      "distillation_loss": 0.042981285601854324,
      "epoch": 6.93,
      "learning_rate": 1.4896486223239802e-08,
      "loss": 0.0405,
      "step": 7295,
      "task_loss": 0.0184940155595541
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997592719204212,
      "compression_loss": 0.0,
      "distillation_loss": 0.02033732831478119,
      "epoch": 6.93,
      "learning_rate": 1.4530977623891662e-08,
      "loss": 0.0188,
      "step": 7296,
      "task_loss": 0.005222789943218231
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997599110211833,
      "compression_loss": 0.0,
      "distillation_loss": 0.03475763648748398,
      "epoch": 6.93,
      "learning_rate": 1.4170007755326398e-08,
      "loss": 0.0357,
      "step": 7297,
      "task_loss": 0.04378265514969826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799760548989792,
      "compression_loss": 0.0,
      "distillation_loss": 0.03861427307128906,
      "epoch": 6.93,
      "learning_rate": 1.3813576683111006e-08,
      "loss": 0.0449,
      "step": 7298,
      "task_loss": 0.10126301646232605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799761185827251,
      "compression_loss": 0.0,
      "distillation_loss": 0.017223935574293137,
      "epoch": 6.93,
      "learning_rate": 1.3461684471993696e-08,
      "loss": 0.0219,
      "step": 7299,
      "task_loss": 0.06419990956783295
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799761821534564,
      "compression_loss": 0.0,
      "distillation_loss": 0.053291283547878265,
      "epoch": 6.93,
      "learning_rate": 1.3114331185898331e-08,
      "loss": 0.0727,
      "step": 7300,
      "task_loss": 0.24761205911636353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997624561127346,
      "compression_loss": 0.0,
      "distillation_loss": 0.0634729415178299,
      "epoch": 6.93,
      "learning_rate": 1.2771516887921664e-08,
      "loss": 0.0634,
      "step": 7301,
      "task_loss": 0.0631205290555954
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997630895627664,
      "compression_loss": 0.0,
      "distillation_loss": 0.018962126225233078,
      "epoch": 6.93,
      "learning_rate": 1.2433241640338878e-08,
      "loss": 0.03,
      "step": 7302,
      "task_loss": 0.12886041402816772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997637218856634,
      "compression_loss": 0.0,
      "distillation_loss": 0.16295194625854492,
      "epoch": 6.94,
      "learning_rate": 1.2099505504600817e-08,
      "loss": 0.169,
      "step": 7303,
      "task_loss": 0.22337132692337036
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799764353082429,
      "compression_loss": 0.0,
      "distillation_loss": 0.027407968416810036,
      "epoch": 6.94,
      "learning_rate": 1.1770308541328434e-08,
      "loss": 0.0343,
      "step": 7304,
      "task_loss": 0.09634220600128174
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997649831540671,
      "compression_loss": 0.0,
      "distillation_loss": 0.023027902469038963,
      "epoch": 6.94,
      "learning_rate": 1.1445650810326668e-08,
      "loss": 0.0246,
      "step": 7305,
      "task_loss": 0.038522087037563324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997656121015811,
      "compression_loss": 0.0,
      "distillation_loss": 0.10117096453905106,
      "epoch": 6.94,
      "learning_rate": 1.1125532370567793e-08,
      "loss": 0.1038,
      "step": 7306,
      "task_loss": 0.12748871743679047
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997662399259748,
      "compression_loss": 0.0,
      "distillation_loss": 0.029107600450515747,
      "epoch": 6.94,
      "learning_rate": 1.0809953280202511e-08,
      "loss": 0.0362,
      "step": 7307,
      "task_loss": 0.09982933849096298
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997668666282521,
      "compression_loss": 0.0,
      "distillation_loss": 0.09483398497104645,
      "epoch": 6.94,
      "learning_rate": 1.0498913596559967e-08,
      "loss": 0.09,
      "step": 7308,
      "task_loss": 0.04666196554899216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997674922094165,
      "compression_loss": 0.0,
      "distillation_loss": 0.025720832869410515,
      "epoch": 6.94,
      "learning_rate": 1.0192413376139408e-08,
      "loss": 0.0305,
      "step": 7309,
      "task_loss": 0.07334055751562119
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997681166704715,
      "compression_loss": 0.0,
      "distillation_loss": 0.0329585038125515,
      "epoch": 6.94,
      "learning_rate": 9.890452674618522e-09,
      "loss": 0.0317,
      "step": 7310,
      "task_loss": 0.020583661273121834
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799768740012421,
      "compression_loss": 0.0,
      "distillation_loss": 0.021503139287233353,
      "epoch": 6.94,
      "learning_rate": 9.593031546853426e-09,
      "loss": 0.0232,
      "step": 7311,
      "task_loss": 0.03804744780063629
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997693622362688,
      "compression_loss": 0.0,
      "distillation_loss": 0.053460683673620224,
      "epoch": 6.94,
      "learning_rate": 9.300150046864796e-09,
      "loss": 0.0505,
      "step": 7312,
      "task_loss": 0.02390519343316555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997699833430184,
      "compression_loss": 0.0,
      "distillation_loss": 0.01604924164712429,
      "epoch": 6.94,
      "learning_rate": 9.011808227865625e-09,
      "loss": 0.0208,
      "step": 7313,
      "task_loss": 0.06401699781417847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997706033336734,
      "compression_loss": 0.0,
      "distillation_loss": 0.012479826807975769,
      "epoch": 6.95,
      "learning_rate": 8.728006142225131e-09,
      "loss": 0.0118,
      "step": 7314,
      "task_loss": 0.005266919732093811
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997712222092377,
      "compression_loss": 0.0,
      "distillation_loss": 0.05143355578184128,
      "epoch": 6.95,
      "learning_rate": 8.448743841504847e-09,
      "loss": 0.0551,
      "step": 7315,
      "task_loss": 0.08826296031475067
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997718399707148,
      "compression_loss": 0.0,
      "distillation_loss": 0.06294449418783188,
      "epoch": 6.95,
      "learning_rate": 8.174021376428087e-09,
      "loss": 0.0601,
      "step": 7316,
      "task_loss": 0.03437443822622299
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997724566191085,
      "compression_loss": 0.0,
      "distillation_loss": 0.08268649876117706,
      "epoch": 6.95,
      "learning_rate": 7.903838796904927e-09,
      "loss": 0.0772,
      "step": 7317,
      "task_loss": 0.028297651559114456
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997730721554224,
      "compression_loss": 0.0,
      "distillation_loss": 0.15433204174041748,
      "epoch": 6.95,
      "learning_rate": 7.638196152010002e-09,
      "loss": 0.1521,
      "step": 7318,
      "task_loss": 0.1320251226425171
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997736865806603,
      "compression_loss": 0.0,
      "distillation_loss": 0.018636398017406464,
      "epoch": 6.95,
      "learning_rate": 7.37709349000193e-09,
      "loss": 0.0172,
      "step": 7319,
      "task_loss": 0.0046669188886880875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997742998958257,
      "compression_loss": 0.0,
      "distillation_loss": 0.0154360830783844,
      "epoch": 6.95,
      "learning_rate": 7.120530858312213e-09,
      "loss": 0.0198,
      "step": 7320,
      "task_loss": 0.05901549756526947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997749121019225,
      "compression_loss": 0.0,
      "distillation_loss": 0.014429114758968353,
      "epoch": 6.95,
      "learning_rate": 6.8685083035452404e-09,
      "loss": 0.014,
      "step": 7321,
      "task_loss": 0.010067546740174294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997755231999542,
      "compression_loss": 0.0,
      "distillation_loss": 0.025817744433879852,
      "epoch": 6.95,
      "learning_rate": 6.621025871481057e-09,
      "loss": 0.0296,
      "step": 7322,
      "task_loss": 0.0636395514011383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997761331909246,
      "compression_loss": 0.0,
      "distillation_loss": 0.033705998212099075,
      "epoch": 6.95,
      "learning_rate": 6.378083607075369e-09,
      "loss": 0.0338,
      "step": 7323,
      "task_loss": 0.03477884456515312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997767420758373,
      "compression_loss": 0.0,
      "distillation_loss": 0.03906092047691345,
      "epoch": 6.96,
      "learning_rate": 6.139681554462318e-09,
      "loss": 0.0507,
      "step": 7324,
      "task_loss": 0.1549626588821411
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997773498556959,
      "compression_loss": 0.0,
      "distillation_loss": 0.051231879740953445,
      "epoch": 6.96,
      "learning_rate": 5.905819756948927e-09,
      "loss": 0.0497,
      "step": 7325,
      "task_loss": 0.03555578738451004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997779565315044,
      "compression_loss": 0.0,
      "distillation_loss": 0.03208434581756592,
      "epoch": 6.96,
      "learning_rate": 5.676498257015106e-09,
      "loss": 0.041,
      "step": 7326,
      "task_loss": 0.12134852260351181
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997785621042662,
      "compression_loss": 0.0,
      "distillation_loss": 0.1095762699842453,
      "epoch": 6.96,
      "learning_rate": 5.4517170963191974e-09,
      "loss": 0.1053,
      "step": 7327,
      "task_loss": 0.06731220334768295
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799779166574985,
      "compression_loss": 0.0,
      "distillation_loss": 0.03222663700580597,
      "epoch": 6.96,
      "learning_rate": 5.231476315695205e-09,
      "loss": 0.0354,
      "step": 7328,
      "task_loss": 0.0638594701886177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997797699446646,
      "compression_loss": 0.0,
      "distillation_loss": 0.01883310079574585,
      "epoch": 6.96,
      "learning_rate": 5.015775955150015e-09,
      "loss": 0.0177,
      "step": 7329,
      "task_loss": 0.007643511518836021
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997803722143086,
      "compression_loss": 0.0,
      "distillation_loss": 0.018376603722572327,
      "epoch": 6.96,
      "learning_rate": 4.804616053866173e-09,
      "loss": 0.0172,
      "step": 7330,
      "task_loss": 0.00702180340886116
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997809733849208,
      "compression_loss": 0.0,
      "distillation_loss": 0.06521350890398026,
      "epoch": 6.96,
      "learning_rate": 4.5979966501991104e-09,
      "loss": 0.0821,
      "step": 7331,
      "task_loss": 0.2336568832397461
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997815734575047,
      "compression_loss": 0.0,
      "distillation_loss": 0.045030295848846436,
      "epoch": 6.96,
      "learning_rate": 4.395917781688242e-09,
      "loss": 0.0426,
      "step": 7332,
      "task_loss": 0.021171823143959045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997821724330642,
      "compression_loss": 0.0,
      "distillation_loss": 0.06834699958562851,
      "epoch": 6.96,
      "learning_rate": 4.19837948503754e-09,
      "loss": 0.0635,
      "step": 7333,
      "task_loss": 0.019571850076317787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997827703126027,
      "compression_loss": 0.0,
      "distillation_loss": 0.039298366755247116,
      "epoch": 6.96,
      "learning_rate": 4.0053817961321905e-09,
      "loss": 0.0437,
      "step": 7334,
      "task_loss": 0.08349810540676117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799783367097124,
      "compression_loss": 0.0,
      "distillation_loss": 0.04764103144407272,
      "epoch": 6.97,
      "learning_rate": 3.8169247500330355e-09,
      "loss": 0.0447,
      "step": 7335,
      "task_loss": 0.01831081695854664
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799783962787632,
      "compression_loss": 0.0,
      "distillation_loss": 0.05844502151012421,
      "epoch": 6.97,
      "learning_rate": 3.633008380971026e-09,
      "loss": 0.0784,
      "step": 7336,
      "task_loss": 0.25827866792678833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997845573851301,
      "compression_loss": 0.0,
      "distillation_loss": 0.03211439028382301,
      "epoch": 6.97,
      "learning_rate": 3.453632722358324e-09,
      "loss": 0.0347,
      "step": 7337,
      "task_loss": 0.058454521000385284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997851508906221,
      "compression_loss": 0.0,
      "distillation_loss": 0.021087612956762314,
      "epoch": 6.97,
      "learning_rate": 3.278797806774425e-09,
      "loss": 0.0199,
      "step": 7338,
      "task_loss": 0.009137500077486038
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997857433051117,
      "compression_loss": 0.0,
      "distillation_loss": 0.07293461263179779,
      "epoch": 6.97,
      "learning_rate": 3.1085036659855847e-09,
      "loss": 0.0815,
      "step": 7339,
      "task_loss": 0.15890468657016754
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997863346296024,
      "compression_loss": 0.0,
      "distillation_loss": 0.04177660495042801,
      "epoch": 6.97,
      "learning_rate": 2.9427503309226166e-09,
      "loss": 0.0499,
      "step": 7340,
      "task_loss": 0.12267521023750305
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997869248650982,
      "compression_loss": 0.0,
      "distillation_loss": 0.06666393578052521,
      "epoch": 6.97,
      "learning_rate": 2.7815378316947694e-09,
      "loss": 0.0758,
      "step": 7341,
      "task_loss": 0.15778383612632751
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997875140126025,
      "compression_loss": 0.0,
      "distillation_loss": 0.04870596528053284,
      "epoch": 6.97,
      "learning_rate": 2.624866197589726e-09,
      "loss": 0.046,
      "step": 7342,
      "task_loss": 0.021872013807296753
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997881020731191,
      "compression_loss": 0.0,
      "distillation_loss": 0.02343062311410904,
      "epoch": 6.97,
      "learning_rate": 2.4727354570680537e-09,
      "loss": 0.0323,
      "step": 7343,
      "task_loss": 0.11193156987428665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997886890476517,
      "compression_loss": 0.0,
      "distillation_loss": 0.0678223967552185,
      "epoch": 6.97,
      "learning_rate": 2.3251456377604288e-09,
      "loss": 0.0726,
      "step": 7344,
      "task_loss": 0.1155007928609848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799789274937204,
      "compression_loss": 0.0,
      "distillation_loss": 0.11146111786365509,
      "epoch": 6.98,
      "learning_rate": 2.1820967664815116e-09,
      "loss": 0.1153,
      "step": 7345,
      "task_loss": 0.15025877952575684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997898597427796,
      "compression_loss": 0.0,
      "distillation_loss": 0.058320388197898865,
      "epoch": 6.98,
      "learning_rate": 2.043588869216073e-09,
      "loss": 0.0701,
      "step": 7346,
      "task_loss": 0.17589399218559265
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997904434653822,
      "compression_loss": 0.0,
      "distillation_loss": 0.03350623697042465,
      "epoch": 6.98,
      "learning_rate": 1.9096219711245423e-09,
      "loss": 0.0307,
      "step": 7347,
      "task_loss": 0.005610005930066109
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997910261060155,
      "compression_loss": 0.0,
      "distillation_loss": 0.02051948755979538,
      "epoch": 6.98,
      "learning_rate": 1.780196096540232e-09,
      "loss": 0.0246,
      "step": 7348,
      "task_loss": 0.06097707524895668
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997916076656832,
      "compression_loss": 0.0,
      "distillation_loss": 0.021704774349927902,
      "epoch": 6.98,
      "learning_rate": 1.6553112689776662e-09,
      "loss": 0.0294,
      "step": 7349,
      "task_loss": 0.0981798768043518
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799792188145389,
      "compression_loss": 0.0,
      "distillation_loss": 0.10756382346153259,
      "epoch": 6.98,
      "learning_rate": 1.5349675111214768e-09,
      "loss": 0.1035,
      "step": 7350,
      "task_loss": 0.06680326163768768
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997927675461366,
      "compression_loss": 0.0,
      "distillation_loss": 0.06751997768878937,
      "epoch": 6.98,
      "learning_rate": 1.419164844831955e-09,
      "loss": 0.0741,
      "step": 7351,
      "task_loss": 0.13287892937660217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997933458689296,
      "compression_loss": 0.0,
      "distillation_loss": 0.028206095099449158,
      "epoch": 6.98,
      "learning_rate": 1.3079032911450516e-09,
      "loss": 0.0329,
      "step": 7352,
      "task_loss": 0.07551902532577515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997939231147716,
      "compression_loss": 0.0,
      "distillation_loss": 0.0961083471775055,
      "epoch": 6.98,
      "learning_rate": 1.201182870272377e-09,
      "loss": 0.0902,
      "step": 7353,
      "task_loss": 0.03751200810074806
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997944992846665,
      "compression_loss": 0.0,
      "distillation_loss": 0.01709740236401558,
      "epoch": 6.98,
      "learning_rate": 1.0990036016012005e-09,
      "loss": 0.0161,
      "step": 7354,
      "task_loss": 0.00756429135799408
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997950743796178,
      "compression_loss": 0.0,
      "distillation_loss": 0.024914680048823357,
      "epoch": 6.98,
      "learning_rate": 1.0013655036916759e-09,
      "loss": 0.0318,
      "step": 7355,
      "task_loss": 0.094090037047863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997956484006293,
      "compression_loss": 0.0,
      "distillation_loss": 0.013773511163890362,
      "epoch": 6.99,
      "learning_rate": 9.082685942796154e-10,
      "loss": 0.0129,
      "step": 7356,
      "task_loss": 0.004718998447060585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997962213487048,
      "compression_loss": 0.0,
      "distillation_loss": 0.020373770967125893,
      "epoch": 6.99,
      "learning_rate": 8.197128902792672e-10,
      "loss": 0.0326,
      "step": 7357,
      "task_loss": 0.1427568942308426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997967932248476,
      "compression_loss": 0.0,
      "distillation_loss": 0.08715569972991943,
      "epoch": 6.99,
      "learning_rate": 7.356984077722117e-10,
      "loss": 0.0883,
      "step": 7358,
      "task_loss": 0.09825003892183304
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997973640300617,
      "compression_loss": 0.0,
      "distillation_loss": 0.03974455967545509,
      "epoch": 6.99,
      "learning_rate": 6.562251620267912e-10,
      "loss": 0.0361,
      "step": 7359,
      "task_loss": 0.0036314092576503754
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997979337653507,
      "compression_loss": 0.0,
      "distillation_loss": 0.10460391640663147,
      "epoch": 6.99,
      "learning_rate": 5.812931674759048e-10,
      "loss": 0.1047,
      "step": 7360,
      "task_loss": 0.10604080557823181
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997985024317182,
      "compression_loss": 0.0,
      "distillation_loss": 0.020527970045804977,
      "epoch": 6.99,
      "learning_rate": 5.109024377308869e-10,
      "loss": 0.0209,
      "step": 7361,
      "task_loss": 0.02422548085451126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799799070030168,
      "compression_loss": 0.0,
      "distillation_loss": 0.09669515490531921,
      "epoch": 6.99,
      "learning_rate": 4.450529855787311e-10,
      "loss": 0.0906,
      "step": 7362,
      "task_loss": 0.03620155528187752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7997996365617038,
      "compression_loss": 0.0,
      "distillation_loss": 0.026915445923805237,
      "epoch": 6.99,
      "learning_rate": 3.8374482298209057e-10,
      "loss": 0.0248,
      "step": 7363,
      "task_loss": 0.005302935838699341
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998002020273292,
      "compression_loss": 0.0,
      "distillation_loss": 0.028947005048394203,
      "epoch": 6.99,
      "learning_rate": 3.2697796107650226e-10,
      "loss": 0.0354,
      "step": 7364,
      "task_loss": 0.09320739656686783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998007664280478,
      "compression_loss": 0.0,
      "distillation_loss": 0.013993775472044945,
      "epoch": 6.99,
      "learning_rate": 2.7475241017871355e-10,
      "loss": 0.0266,
      "step": 7365,
      "task_loss": 0.1398152858018875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998013297648634,
      "compression_loss": 0.0,
      "distillation_loss": 0.0770372524857521,
      "epoch": 7.0,
      "learning_rate": 2.2706817977002914e-10,
      "loss": 0.0781,
      "step": 7366,
      "task_loss": 0.0880659967660904
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998018920387797,
      "compression_loss": 0.0,
      "distillation_loss": 0.052641429007053375,
      "epoch": 7.0,
      "learning_rate": 1.8392527851296415e-10,
      "loss": 0.0551,
      "step": 7367,
      "task_loss": 0.0774116963148117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998024532508003,
      "compression_loss": 0.0,
      "distillation_loss": 0.05743606388568878,
      "epoch": 7.0,
      "learning_rate": 1.453237142484687e-10,
      "loss": 0.0612,
      "step": 7368,
      "task_loss": 0.09504348039627075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799803013401929,
      "compression_loss": 0.0,
      "distillation_loss": 0.026639528572559357,
      "epoch": 7.0,
      "learning_rate": 1.112634939848256e-10,
      "loss": 0.0321,
      "step": 7369,
      "task_loss": 0.08151736855506897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998035724931695,
      "compression_loss": 0.0,
      "distillation_loss": 0.03500088304281235,
      "epoch": 7.0,
      "learning_rate": 8.174462391430382e-11,
      "loss": 0.0384,
      "step": 7370,
      "task_loss": 0.06867900490760803
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6237196162521903,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7984375187077579,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998041305255252,
      "compression_loss": 0.0,
      "distillation_loss": 0.012465332634747028,
      "epoch": 7.0,
      "learning_rate": 5.676710939372942e-11,
      "loss": 0.0127,
      "step": 7371,
      "task_loss": 0.014598931185901165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998046875,
      "compression_loss": 0.0,
      "distillation_loss": 0.09104140102863312,
      "epoch": 7.0,
      "learning_rate": 3.633095496113903e-11,
      "loss": 0.0992,
      "step": 7372,
      "task_loss": 0.17217549681663513
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998052434175976,
      "compression_loss": 0.0,
      "distillation_loss": 0.07841981947422028,
      "epoch": 7.0,
      "learning_rate": 2.0436164330228658e-11,
      "loss": 0.0708,
      "step": 7373,
      "task_loss": 0.0018375962972640991
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998057982793217,
      "compression_loss": 0.0,
      "distillation_loss": 0.08232280611991882,
      "epoch": 7.0,
      "learning_rate": 9.082740390353728e-12,
      "loss": 0.0767,
      "step": 7374,
      "task_loss": 0.026533475145697594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998063520861759,
      "compression_loss": 0.0,
      "distillation_loss": 0.07200402021408081,
      "epoch": 7.0,
      "learning_rate": 2.270685200977951e-12,
      "loss": 0.0776,
      "step": 7375,
      "task_loss": 0.12824085354804993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998069048391638,
      "compression_loss": 0.0,
      "distillation_loss": 0.19095724821090698,
      "epoch": 7.0,
      "learning_rate": 5e-05,
      "loss": 0.1839,
      "step": 7376,
      "task_loss": 0.12048640847206116
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998074565392893,
      "compression_loss": 0.0,
      "distillation_loss": 0.04641976207494736,
      "epoch": 7.01,
      "learning_rate": 4.99999977293148e-05,
      "loss": 0.05,
      "step": 7377,
      "task_loss": 0.0825609490275383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998080071875558,
      "compression_loss": 0.0,
      "distillation_loss": 0.1098412275314331,
      "epoch": 7.01,
      "learning_rate": 4.999999091725961e-05,
      "loss": 0.111,
      "step": 7378,
      "task_loss": 0.12164577841758728
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998085567849673,
      "compression_loss": 0.0,
      "distillation_loss": 0.07424164563417435,
      "epoch": 7.01,
      "learning_rate": 4.999997956383567e-05,
      "loss": 0.0833,
      "step": 7379,
      "task_loss": 0.16513806581497192
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998091053325274,
      "compression_loss": 0.0,
      "distillation_loss": 0.08631199598312378,
      "epoch": 7.01,
      "learning_rate": 4.999996366904504e-05,
      "loss": 0.0779,
      "step": 7380,
      "task_loss": 0.002591833472251892
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998096528312395,
      "compression_loss": 0.0,
      "distillation_loss": 0.09567269682884216,
      "epoch": 7.01,
      "learning_rate": 4.999994323289061e-05,
      "loss": 0.0913,
      "step": 7381,
      "task_loss": 0.05151829496026039
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998101992821076,
      "compression_loss": 0.0,
      "distillation_loss": 0.0753033235669136,
      "epoch": 7.01,
      "learning_rate": 4.999991825537609e-05,
      "loss": 0.0694,
      "step": 7382,
      "task_loss": 0.016505436971783638
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998107446861353,
      "compression_loss": 0.0,
      "distillation_loss": 0.038341403007507324,
      "epoch": 7.01,
      "learning_rate": 4.999988873650602e-05,
      "loss": 0.0354,
      "step": 7383,
      "task_loss": 0.008481014519929886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998112890443263,
      "compression_loss": 0.0,
      "distillation_loss": 0.04590607061982155,
      "epoch": 7.01,
      "learning_rate": 4.999985467628575e-05,
      "loss": 0.0532,
      "step": 7384,
      "task_loss": 0.11850175261497498
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998118323576842,
      "compression_loss": 0.0,
      "distillation_loss": 0.06865381449460983,
      "epoch": 7.01,
      "learning_rate": 4.999981607472149e-05,
      "loss": 0.0665,
      "step": 7385,
      "task_loss": 0.046819057315588
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998123746272127,
      "compression_loss": 0.0,
      "distillation_loss": 0.06713886559009552,
      "epoch": 7.01,
      "learning_rate": 4.999977293182023e-05,
      "loss": 0.0622,
      "step": 7386,
      "task_loss": 0.018027808517217636
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998129158539155,
      "compression_loss": 0.0,
      "distillation_loss": 0.03523305803537369,
      "epoch": 7.02,
      "learning_rate": 4.999972524758982e-05,
      "loss": 0.0459,
      "step": 7387,
      "task_loss": 0.14175420999526978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998134560387963,
      "compression_loss": 0.0,
      "distillation_loss": 0.10418183356523514,
      "epoch": 7.02,
      "learning_rate": 4.999967302203893e-05,
      "loss": 0.106,
      "step": 7388,
      "task_loss": 0.12198633700609207
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998139951828587,
      "compression_loss": 0.0,
      "distillation_loss": 0.02372100204229355,
      "epoch": 7.02,
      "learning_rate": 4.9999616255177016e-05,
      "loss": 0.0308,
      "step": 7389,
      "task_loss": 0.09421825408935547
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998145332871065,
      "compression_loss": 0.0,
      "distillation_loss": 0.028630398213863373,
      "epoch": 7.02,
      "learning_rate": 4.999955494701443e-05,
      "loss": 0.0346,
      "step": 7390,
      "task_loss": 0.08789223432540894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998150703525434,
      "compression_loss": 0.0,
      "distillation_loss": 0.14551189541816711,
      "epoch": 7.02,
      "learning_rate": 4.999948909756227e-05,
      "loss": 0.1518,
      "step": 7391,
      "task_loss": 0.20824161171913147
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998156063801729,
      "compression_loss": 0.0,
      "distillation_loss": 0.13446444272994995,
      "epoch": 7.02,
      "learning_rate": 4.9999418706832525e-05,
      "loss": 0.1363,
      "step": 7392,
      "task_loss": 0.15303920209407806
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998161413709989,
      "compression_loss": 0.0,
      "distillation_loss": 0.025591343641281128,
      "epoch": 7.02,
      "learning_rate": 4.9999343774837976e-05,
      "loss": 0.0238,
      "step": 7393,
      "task_loss": 0.007202155888080597
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799816675326025,
      "compression_loss": 0.0,
      "distillation_loss": 0.03282269090414047,
      "epoch": 7.02,
      "learning_rate": 4.999926430159223e-05,
      "loss": 0.031,
      "step": 7394,
      "task_loss": 0.014277882874011993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998172082462548,
      "compression_loss": 0.0,
      "distillation_loss": 0.05018794909119606,
      "epoch": 7.02,
      "learning_rate": 4.9999180287109725e-05,
      "loss": 0.0654,
      "step": 7395,
      "task_loss": 0.20274071395397186
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799817740132692,
      "compression_loss": 0.0,
      "distillation_loss": 0.027372337877750397,
      "epoch": 7.02,
      "learning_rate": 4.999909173140572e-05,
      "loss": 0.0256,
      "step": 7396,
      "task_loss": 0.01010635495185852
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998182709863404,
      "compression_loss": 0.0,
      "distillation_loss": 0.02192852459847927,
      "epoch": 7.02,
      "learning_rate": 4.999899863449631e-05,
      "loss": 0.0256,
      "step": 7397,
      "task_loss": 0.05868425592780113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998188008082036,
      "compression_loss": 0.0,
      "distillation_loss": 0.06173873692750931,
      "epoch": 7.03,
      "learning_rate": 4.99989009963984e-05,
      "loss": 0.0675,
      "step": 7398,
      "task_loss": 0.1194574236869812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998193295992853,
      "compression_loss": 0.0,
      "distillation_loss": 0.06959561258554459,
      "epoch": 7.03,
      "learning_rate": 4.999879881712973e-05,
      "loss": 0.0697,
      "step": 7399,
      "task_loss": 0.07095710933208466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998198573605891,
      "compression_loss": 0.0,
      "distillation_loss": 0.08823947608470917,
      "epoch": 7.03,
      "learning_rate": 4.999869209670885e-05,
      "loss": 0.0889,
      "step": 7400,
      "task_loss": 0.0944376289844513
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998203840931188,
      "compression_loss": 0.0,
      "distillation_loss": 0.03558321297168732,
      "epoch": 7.03,
      "learning_rate": 4.999858083515517e-05,
      "loss": 0.0466,
      "step": 7401,
      "task_loss": 0.14564195275306702
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799820909797878,
      "compression_loss": 0.0,
      "distillation_loss": 0.1047942191362381,
      "epoch": 7.03,
      "learning_rate": 4.999846503248888e-05,
      "loss": 0.1106,
      "step": 7402,
      "task_loss": 0.16298425197601318
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998214344758705,
      "compression_loss": 0.0,
      "distillation_loss": 0.03409132733941078,
      "epoch": 7.03,
      "learning_rate": 4.9998344688731027e-05,
      "loss": 0.0435,
      "step": 7403,
      "task_loss": 0.12837141752243042
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998219581280999,
      "compression_loss": 0.0,
      "distillation_loss": 0.025968341156840324,
      "epoch": 7.03,
      "learning_rate": 4.999821980390346e-05,
      "loss": 0.0237,
      "step": 7404,
      "task_loss": 0.003702618181705475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998224807555698,
      "compression_loss": 0.0,
      "distillation_loss": 0.04202386364340782,
      "epoch": 7.03,
      "learning_rate": 4.999809037802888e-05,
      "loss": 0.0391,
      "step": 7405,
      "task_loss": 0.013203632086515427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799823002359284,
      "compression_loss": 0.0,
      "distillation_loss": 0.020798005163669586,
      "epoch": 7.03,
      "learning_rate": 4.999795641113079e-05,
      "loss": 0.0203,
      "step": 7406,
      "task_loss": 0.016108643263578415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998235229402462,
      "compression_loss": 0.0,
      "distillation_loss": 0.04067230597138405,
      "epoch": 7.03,
      "learning_rate": 4.9997817903233527e-05,
      "loss": 0.0412,
      "step": 7407,
      "task_loss": 0.04560984671115875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998240424994599,
      "compression_loss": 0.0,
      "distillation_loss": 0.03840542584657669,
      "epoch": 7.04,
      "learning_rate": 4.999767485436224e-05,
      "loss": 0.0402,
      "step": 7408,
      "task_loss": 0.05598912388086319
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998245610379291,
      "compression_loss": 0.0,
      "distillation_loss": 0.028701873496174812,
      "epoch": 7.04,
      "learning_rate": 4.999752726454293e-05,
      "loss": 0.0274,
      "step": 7409,
      "task_loss": 0.015496550127863884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998250785566572,
      "compression_loss": 0.0,
      "distillation_loss": 0.03353622183203697,
      "epoch": 7.04,
      "learning_rate": 4.9997375133802415e-05,
      "loss": 0.0376,
      "step": 7410,
      "task_loss": 0.07444252073764801
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799825595056648,
      "compression_loss": 0.0,
      "distillation_loss": 0.021153662353754044,
      "epoch": 7.04,
      "learning_rate": 4.999721846216831e-05,
      "loss": 0.0196,
      "step": 7411,
      "task_loss": 0.005829468369483948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998261105389051,
      "compression_loss": 0.0,
      "distillation_loss": 0.059076711535453796,
      "epoch": 7.04,
      "learning_rate": 4.999705724966908e-05,
      "loss": 0.0687,
      "step": 7412,
      "task_loss": 0.15493744611740112
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998266250044322,
      "compression_loss": 0.0,
      "distillation_loss": 0.06167703866958618,
      "epoch": 7.04,
      "learning_rate": 4.999689149633402e-05,
      "loss": 0.0608,
      "step": 7413,
      "task_loss": 0.05248799920082092
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799827138454233,
      "compression_loss": 0.0,
      "distillation_loss": 0.02228119783103466,
      "epoch": 7.04,
      "learning_rate": 4.999672120219323e-05,
      "loss": 0.0276,
      "step": 7414,
      "task_loss": 0.07509797811508179
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998276508893114,
      "compression_loss": 0.0,
      "distillation_loss": 0.03827614709734917,
      "epoch": 7.04,
      "learning_rate": 4.999654636727764e-05,
      "loss": 0.0452,
      "step": 7415,
      "task_loss": 0.10720621794462204
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998281623106708,
      "compression_loss": 0.0,
      "distillation_loss": 0.10008041560649872,
      "epoch": 7.04,
      "learning_rate": 4.9996366991619034e-05,
      "loss": 0.0968,
      "step": 7416,
      "task_loss": 0.06756041944026947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998286727193149,
      "compression_loss": 0.0,
      "distillation_loss": 0.04735985025763512,
      "epoch": 7.04,
      "learning_rate": 4.999618307524997e-05,
      "loss": 0.0456,
      "step": 7417,
      "task_loss": 0.02975446730852127
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998291821162474,
      "compression_loss": 0.0,
      "distillation_loss": 0.02187761291861534,
      "epoch": 7.04,
      "learning_rate": 4.999599461820387e-05,
      "loss": 0.0201,
      "step": 7418,
      "task_loss": 0.0042492058128118515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998296905024722,
      "compression_loss": 0.0,
      "distillation_loss": 0.020126711577177048,
      "epoch": 7.05,
      "learning_rate": 4.999580162051497e-05,
      "loss": 0.0187,
      "step": 7419,
      "task_loss": 0.005694573745131493
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998301978789928,
      "compression_loss": 0.0,
      "distillation_loss": 0.02660163678228855,
      "epoch": 7.05,
      "learning_rate": 4.9995604082218314e-05,
      "loss": 0.0327,
      "step": 7420,
      "task_loss": 0.08736631274223328
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998307042468128,
      "compression_loss": 0.0,
      "distillation_loss": 0.05452951416373253,
      "epoch": 7.05,
      "learning_rate": 4.99954020033498e-05,
      "loss": 0.0549,
      "step": 7421,
      "task_loss": 0.058631766587495804
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799831209606936,
      "compression_loss": 0.0,
      "distillation_loss": 0.0896400734782219,
      "epoch": 7.05,
      "learning_rate": 4.9995195383946135e-05,
      "loss": 0.0861,
      "step": 7422,
      "task_loss": 0.053857218474149704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998317139603661,
      "compression_loss": 0.0,
      "distillation_loss": 0.026753665879368782,
      "epoch": 7.05,
      "learning_rate": 4.999498422404485e-05,
      "loss": 0.0279,
      "step": 7423,
      "task_loss": 0.03794853016734123
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998322173081067,
      "compression_loss": 0.0,
      "distillation_loss": 0.023596402257680893,
      "epoch": 7.05,
      "learning_rate": 4.999476852368431e-05,
      "loss": 0.0243,
      "step": 7424,
      "task_loss": 0.030923640355467796
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998327196511615,
      "compression_loss": 0.0,
      "distillation_loss": 0.060296621173620224,
      "epoch": 7.05,
      "learning_rate": 4.999454828290369e-05,
      "loss": 0.0597,
      "step": 7425,
      "task_loss": 0.05454102158546448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998332209905342,
      "compression_loss": 0.0,
      "distillation_loss": 0.0592413991689682,
      "epoch": 7.05,
      "learning_rate": 4.999432350174299e-05,
      "loss": 0.0575,
      "step": 7426,
      "task_loss": 0.04156405106186867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998337213272286,
      "compression_loss": 0.0,
      "distillation_loss": 0.07152185589075089,
      "epoch": 7.05,
      "learning_rate": 4.9994094180243055e-05,
      "loss": 0.0712,
      "step": 7427,
      "task_loss": 0.0684933215379715
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998342206622482,
      "compression_loss": 0.0,
      "distillation_loss": 0.0811118632555008,
      "epoch": 7.05,
      "learning_rate": 4.999386031844554e-05,
      "loss": 0.0814,
      "step": 7428,
      "task_loss": 0.08375652134418488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998347189965967,
      "compression_loss": 0.0,
      "distillation_loss": 0.0416887030005455,
      "epoch": 7.06,
      "learning_rate": 4.999362191639293e-05,
      "loss": 0.0431,
      "step": 7429,
      "task_loss": 0.05558731034398079
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799835216331278,
      "compression_loss": 0.0,
      "distillation_loss": 0.04246686398983002,
      "epoch": 7.06,
      "learning_rate": 4.999337897412852e-05,
      "loss": 0.0439,
      "step": 7430,
      "task_loss": 0.05704366788268089
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998357126672955,
      "compression_loss": 0.0,
      "distillation_loss": 0.028290575370192528,
      "epoch": 7.06,
      "learning_rate": 4.999313149169645e-05,
      "loss": 0.0431,
      "step": 7431,
      "task_loss": 0.17606376111507416
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799836208005653,
      "compression_loss": 0.0,
      "distillation_loss": 0.08979344367980957,
      "epoch": 7.06,
      "learning_rate": 4.999287946914169e-05,
      "loss": 0.0868,
      "step": 7432,
      "task_loss": 0.06011103093624115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998367023473542,
      "compression_loss": 0.0,
      "distillation_loss": 0.06186497583985329,
      "epoch": 7.06,
      "learning_rate": 4.999262290651e-05,
      "loss": 0.0586,
      "step": 7433,
      "task_loss": 0.02930166944861412
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998371956934027,
      "compression_loss": 0.0,
      "distillation_loss": 0.022470489144325256,
      "epoch": 7.06,
      "learning_rate": 4.9992361803847995e-05,
      "loss": 0.0208,
      "step": 7434,
      "task_loss": 0.005419734865427017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998376880448024,
      "compression_loss": 0.0,
      "distillation_loss": 0.056562915444374084,
      "epoch": 7.06,
      "learning_rate": 4.99920961612031e-05,
      "loss": 0.0685,
      "step": 7435,
      "task_loss": 0.17543764412403107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998381794025567,
      "compression_loss": 0.0,
      "distillation_loss": 0.05138598382472992,
      "epoch": 7.06,
      "learning_rate": 4.9991825978623574e-05,
      "loss": 0.0542,
      "step": 7436,
      "task_loss": 0.07962413132190704
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998386697676695,
      "compression_loss": 0.0,
      "distillation_loss": 0.015319476835429668,
      "epoch": 7.06,
      "learning_rate": 4.9991551256158495e-05,
      "loss": 0.0277,
      "step": 7437,
      "task_loss": 0.13944551348686218
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998391591411443,
      "compression_loss": 0.0,
      "distillation_loss": 0.01975143700838089,
      "epoch": 7.06,
      "learning_rate": 4.999127199385778e-05,
      "loss": 0.0257,
      "step": 7438,
      "task_loss": 0.07912784814834595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799839647523985,
      "compression_loss": 0.0,
      "distillation_loss": 0.07420433312654495,
      "epoch": 7.06,
      "learning_rate": 4.999098819177214e-05,
      "loss": 0.0816,
      "step": 7439,
      "task_loss": 0.14786875247955322
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799840134917195,
      "compression_loss": 0.0,
      "distillation_loss": 0.027095727622509003,
      "epoch": 7.07,
      "learning_rate": 4.999069984995314e-05,
      "loss": 0.0319,
      "step": 7440,
      "task_loss": 0.07478499412536621
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998406213217781,
      "compression_loss": 0.0,
      "distillation_loss": 0.04266301542520523,
      "epoch": 7.07,
      "learning_rate": 4.999040696845315e-05,
      "loss": 0.0465,
      "step": 7441,
      "task_loss": 0.08147358894348145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998411067387382,
      "compression_loss": 0.0,
      "distillation_loss": 0.021399151533842087,
      "epoch": 7.07,
      "learning_rate": 4.999010954732538e-05,
      "loss": 0.026,
      "step": 7442,
      "task_loss": 0.06721732020378113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998415911690787,
      "compression_loss": 0.0,
      "distillation_loss": 0.04915117844939232,
      "epoch": 7.07,
      "learning_rate": 4.998980758662386e-05,
      "loss": 0.0452,
      "step": 7443,
      "task_loss": 0.009514054283499718
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998420746138034,
      "compression_loss": 0.0,
      "distillation_loss": 0.08443302661180496,
      "epoch": 7.07,
      "learning_rate": 4.998950108640345e-05,
      "loss": 0.0882,
      "step": 7444,
      "task_loss": 0.12241680175065994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799842557073916,
      "compression_loss": 0.0,
      "distillation_loss": 0.03232605382800102,
      "epoch": 7.07,
      "learning_rate": 4.99891900467198e-05,
      "loss": 0.0313,
      "step": 7445,
      "task_loss": 0.022218016907572746
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998430385504202,
      "compression_loss": 0.0,
      "distillation_loss": 0.032280419021844864,
      "epoch": 7.07,
      "learning_rate": 4.9988874467629435e-05,
      "loss": 0.0345,
      "step": 7446,
      "task_loss": 0.05450471490621567
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998435190443195,
      "compression_loss": 0.0,
      "distillation_loss": 0.027930978685617447,
      "epoch": 7.07,
      "learning_rate": 4.998855434918968e-05,
      "loss": 0.0262,
      "step": 7447,
      "task_loss": 0.010617373511195183
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998439985566178,
      "compression_loss": 0.0,
      "distillation_loss": 0.06132368743419647,
      "epoch": 7.07,
      "learning_rate": 4.998822969145868e-05,
      "loss": 0.0739,
      "step": 7448,
      "task_loss": 0.18746685981750488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998444770883186,
      "compression_loss": 0.0,
      "distillation_loss": 0.03330961614847183,
      "epoch": 7.07,
      "learning_rate": 4.99879004944954e-05,
      "loss": 0.0313,
      "step": 7449,
      "task_loss": 0.012728353962302208
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998449546404258,
      "compression_loss": 0.0,
      "distillation_loss": 0.018877487629652023,
      "epoch": 7.08,
      "learning_rate": 4.998756675835966e-05,
      "loss": 0.0173,
      "step": 7450,
      "task_loss": 0.0031468812376260757
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998454312139429,
      "compression_loss": 0.0,
      "distillation_loss": 0.08477997779846191,
      "epoch": 7.08,
      "learning_rate": 4.9987228483112083e-05,
      "loss": 0.0814,
      "step": 7451,
      "task_loss": 0.05088125914335251
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998459068098737,
      "compression_loss": 0.0,
      "distillation_loss": 0.05044269561767578,
      "epoch": 7.08,
      "learning_rate": 4.998688566881411e-05,
      "loss": 0.0471,
      "step": 7452,
      "task_loss": 0.016654808074235916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998463814292217,
      "compression_loss": 0.0,
      "distillation_loss": 0.04332347959280014,
      "epoch": 7.08,
      "learning_rate": 4.998653831552801e-05,
      "loss": 0.0424,
      "step": 7453,
      "task_loss": 0.034489430487155914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998468550729908,
      "compression_loss": 0.0,
      "distillation_loss": 0.09742948412895203,
      "epoch": 7.08,
      "learning_rate": 4.998618642331689e-05,
      "loss": 0.1015,
      "step": 7454,
      "task_loss": 0.13810521364212036
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998473277421846,
      "compression_loss": 0.0,
      "distillation_loss": 0.06396178901195526,
      "epoch": 7.08,
      "learning_rate": 4.9985829992244675e-05,
      "loss": 0.0684,
      "step": 7455,
      "task_loss": 0.10790328681468964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998477994378067,
      "compression_loss": 0.0,
      "distillation_loss": 0.04675624519586563,
      "epoch": 7.08,
      "learning_rate": 4.998546902237611e-05,
      "loss": 0.0458,
      "step": 7456,
      "task_loss": 0.03740784153342247
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998482701608609,
      "compression_loss": 0.0,
      "distillation_loss": 0.018789753317832947,
      "epoch": 7.08,
      "learning_rate": 4.9985103513776764e-05,
      "loss": 0.0245,
      "step": 7457,
      "task_loss": 0.07571886479854584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998487399123507,
      "compression_loss": 0.0,
      "distillation_loss": 0.06566867977380753,
      "epoch": 7.08,
      "learning_rate": 4.998473346651303e-05,
      "loss": 0.0631,
      "step": 7458,
      "task_loss": 0.03988581523299217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998492086932801,
      "compression_loss": 0.0,
      "distillation_loss": 0.03311315178871155,
      "epoch": 7.08,
      "learning_rate": 4.9984358880652146e-05,
      "loss": 0.0408,
      "step": 7459,
      "task_loss": 0.1103094071149826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998496765046526,
      "compression_loss": 0.0,
      "distillation_loss": 0.11287416517734528,
      "epoch": 7.08,
      "learning_rate": 4.9983979756262136e-05,
      "loss": 0.1219,
      "step": 7460,
      "task_loss": 0.20322994887828827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998501433474717,
      "compression_loss": 0.0,
      "distillation_loss": 0.026843059808015823,
      "epoch": 7.09,
      "learning_rate": 4.998359609341188e-05,
      "loss": 0.037,
      "step": 7461,
      "task_loss": 0.12811830639839172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998506092227414,
      "compression_loss": 0.0,
      "distillation_loss": 0.06953533738851547,
      "epoch": 7.09,
      "learning_rate": 4.9983207892171074e-05,
      "loss": 0.0673,
      "step": 7462,
      "task_loss": 0.04702718183398247
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998510741314653,
      "compression_loss": 0.0,
      "distillation_loss": 0.05555496737360954,
      "epoch": 7.09,
      "learning_rate": 4.998281515261023e-05,
      "loss": 0.0607,
      "step": 7463,
      "task_loss": 0.10690201073884964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998515380746469,
      "compression_loss": 0.0,
      "distillation_loss": 0.025969993323087692,
      "epoch": 7.09,
      "learning_rate": 4.9982417874800704e-05,
      "loss": 0.0266,
      "step": 7464,
      "task_loss": 0.03247838839888573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79985200105329,
      "compression_loss": 0.0,
      "distillation_loss": 0.057906825095415115,
      "epoch": 7.09,
      "learning_rate": 4.998201605881465e-05,
      "loss": 0.0635,
      "step": 7465,
      "task_loss": 0.1140889823436737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998524630683984,
      "compression_loss": 0.0,
      "distillation_loss": 0.019549962133169174,
      "epoch": 7.09,
      "learning_rate": 4.9981609704725057e-05,
      "loss": 0.0261,
      "step": 7466,
      "task_loss": 0.08523699641227722
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998529241209756,
      "compression_loss": 0.0,
      "distillation_loss": 0.14493508636951447,
      "epoch": 7.09,
      "learning_rate": 4.998119881260576e-05,
      "loss": 0.1549,
      "step": 7467,
      "task_loss": 0.244978129863739
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998533842120253,
      "compression_loss": 0.0,
      "distillation_loss": 0.10177647322416306,
      "epoch": 7.09,
      "learning_rate": 4.9980783382531376e-05,
      "loss": 0.1153,
      "step": 7468,
      "task_loss": 0.23712855577468872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998538433425514,
      "compression_loss": 0.0,
      "distillation_loss": 0.0323081910610199,
      "epoch": 7.09,
      "learning_rate": 4.998036341457739e-05,
      "loss": 0.0372,
      "step": 7469,
      "task_loss": 0.08089584857225418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998543015135573,
      "compression_loss": 0.0,
      "distillation_loss": 0.045823805034160614,
      "epoch": 7.09,
      "learning_rate": 4.997993890882008e-05,
      "loss": 0.0499,
      "step": 7470,
      "task_loss": 0.08696474879980087
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998547587260468,
      "compression_loss": 0.0,
      "distillation_loss": 0.03767596185207367,
      "epoch": 7.09,
      "learning_rate": 4.997950986533656e-05,
      "loss": 0.0433,
      "step": 7471,
      "task_loss": 0.0938701331615448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998552149810236,
      "compression_loss": 0.0,
      "distillation_loss": 0.040337882936000824,
      "epoch": 7.1,
      "learning_rate": 4.997907628420477e-05,
      "loss": 0.0532,
      "step": 7472,
      "task_loss": 0.16945821046829224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998556702794914,
      "compression_loss": 0.0,
      "distillation_loss": 0.0397014319896698,
      "epoch": 7.1,
      "learning_rate": 4.9978638165503475e-05,
      "loss": 0.0405,
      "step": 7473,
      "task_loss": 0.04727710038423538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998561246224538,
      "compression_loss": 0.0,
      "distillation_loss": 0.04514710605144501,
      "epoch": 7.1,
      "learning_rate": 4.9978195509312266e-05,
      "loss": 0.061,
      "step": 7474,
      "task_loss": 0.20394141972064972
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998565780109146,
      "compression_loss": 0.0,
      "distillation_loss": 0.13967446982860565,
      "epoch": 7.1,
      "learning_rate": 4.997774831571154e-05,
      "loss": 0.143,
      "step": 7475,
      "task_loss": 0.17266049981117249
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998570304458774,
      "compression_loss": 0.0,
      "distillation_loss": 0.043175503611564636,
      "epoch": 7.1,
      "learning_rate": 4.9977296584782544e-05,
      "loss": 0.0402,
      "step": 7476,
      "task_loss": 0.013554053381085396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998574819283458,
      "compression_loss": 0.0,
      "distillation_loss": 0.024114977568387985,
      "epoch": 7.1,
      "learning_rate": 4.997684031660732e-05,
      "loss": 0.0293,
      "step": 7477,
      "task_loss": 0.07583250850439072
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998579324593237,
      "compression_loss": 0.0,
      "distillation_loss": 0.04103930667042732,
      "epoch": 7.1,
      "learning_rate": 4.997637951126877e-05,
      "loss": 0.0476,
      "step": 7478,
      "task_loss": 0.1067778468132019
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998583820398146,
      "compression_loss": 0.0,
      "distillation_loss": 0.10217143595218658,
      "epoch": 7.1,
      "learning_rate": 4.997591416885059e-05,
      "loss": 0.1005,
      "step": 7479,
      "task_loss": 0.08553251624107361
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998588306708222,
      "compression_loss": 0.0,
      "distillation_loss": 0.021832741796970367,
      "epoch": 7.1,
      "learning_rate": 4.997544428943732e-05,
      "loss": 0.0202,
      "step": 7480,
      "task_loss": 0.005031948909163475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998592783533502,
      "compression_loss": 0.0,
      "distillation_loss": 0.024123501032590866,
      "epoch": 7.1,
      "learning_rate": 4.997496987311431e-05,
      "loss": 0.0222,
      "step": 7481,
      "task_loss": 0.0046864766627550125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998597250884024,
      "compression_loss": 0.0,
      "distillation_loss": 0.01681506633758545,
      "epoch": 7.11,
      "learning_rate": 4.997449091996774e-05,
      "loss": 0.023,
      "step": 7482,
      "task_loss": 0.0789770558476448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998601708769824,
      "compression_loss": 0.0,
      "distillation_loss": 0.031241536140441895,
      "epoch": 7.11,
      "learning_rate": 4.9974007430084617e-05,
      "loss": 0.0407,
      "step": 7483,
      "task_loss": 0.12625262141227722
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998606157200938,
      "compression_loss": 0.0,
      "distillation_loss": 0.05506356433033943,
      "epoch": 7.11,
      "learning_rate": 4.997351940355277e-05,
      "loss": 0.0503,
      "step": 7484,
      "task_loss": 0.0069837626069784164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998610596187403,
      "compression_loss": 0.0,
      "distillation_loss": 0.02426174283027649,
      "epoch": 7.11,
      "learning_rate": 4.997302684046085e-05,
      "loss": 0.0417,
      "step": 7485,
      "task_loss": 0.19815370440483093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998615025739257,
      "compression_loss": 0.0,
      "distillation_loss": 0.014478189870715141,
      "epoch": 7.11,
      "learning_rate": 4.997252974089833e-05,
      "loss": 0.0135,
      "step": 7486,
      "task_loss": 0.004255037754774094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998619445866537,
      "compression_loss": 0.0,
      "distillation_loss": 0.03041962906718254,
      "epoch": 7.11,
      "learning_rate": 4.997202810495551e-05,
      "loss": 0.0277,
      "step": 7487,
      "task_loss": 0.0034227408468723297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998623856579278,
      "compression_loss": 0.0,
      "distillation_loss": 0.04478137940168381,
      "epoch": 7.11,
      "learning_rate": 4.997152193272353e-05,
      "loss": 0.0427,
      "step": 7488,
      "task_loss": 0.024202093482017517
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998628257887518,
      "compression_loss": 0.0,
      "distillation_loss": 0.016431665048003197,
      "epoch": 7.11,
      "learning_rate": 4.9971011224294314e-05,
      "loss": 0.0153,
      "step": 7489,
      "task_loss": 0.004669001325964928
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998632649801293,
      "compression_loss": 0.0,
      "distillation_loss": 0.042685117572546005,
      "epoch": 7.11,
      "learning_rate": 4.997049597976066e-05,
      "loss": 0.0468,
      "step": 7490,
      "task_loss": 0.0839599221944809
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998637032330641,
      "compression_loss": 0.0,
      "distillation_loss": 0.038373369723558426,
      "epoch": 7.11,
      "learning_rate": 4.9969976199216144e-05,
      "loss": 0.0551,
      "step": 7491,
      "task_loss": 0.2055097371339798
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998641405485598,
      "compression_loss": 0.0,
      "distillation_loss": 0.01765899360179901,
      "epoch": 7.11,
      "learning_rate": 4.9969451882755196e-05,
      "loss": 0.0163,
      "step": 7492,
      "task_loss": 0.0037982575595378876
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998645769276201,
      "compression_loss": 0.0,
      "distillation_loss": 0.019907163456082344,
      "epoch": 7.12,
      "learning_rate": 4.996892303047306e-05,
      "loss": 0.0281,
      "step": 7493,
      "task_loss": 0.10134420543909073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998650123712487,
      "compression_loss": 0.0,
      "distillation_loss": 0.02686948888003826,
      "epoch": 7.12,
      "learning_rate": 4.996838964246581e-05,
      "loss": 0.0298,
      "step": 7494,
      "task_loss": 0.055808331817388535
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998654468804494,
      "compression_loss": 0.0,
      "distillation_loss": 0.07542771100997925,
      "epoch": 7.12,
      "learning_rate": 4.996785171883032e-05,
      "loss": 0.0714,
      "step": 7495,
      "task_loss": 0.034990422427654266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998658804562255,
      "compression_loss": 0.0,
      "distillation_loss": 0.026508131995797157,
      "epoch": 7.12,
      "learning_rate": 4.996730925966433e-05,
      "loss": 0.0363,
      "step": 7496,
      "task_loss": 0.12431460618972778
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998663130995811,
      "compression_loss": 0.0,
      "distillation_loss": 0.08525696396827698,
      "epoch": 7.12,
      "learning_rate": 4.996676226506636e-05,
      "loss": 0.0831,
      "step": 7497,
      "task_loss": 0.06366419792175293
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998667448115198,
      "compression_loss": 0.0,
      "distillation_loss": 0.02573692798614502,
      "epoch": 7.12,
      "learning_rate": 4.9966210735135785e-05,
      "loss": 0.0237,
      "step": 7498,
      "task_loss": 0.005860496312379837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799867175593045,
      "compression_loss": 0.0,
      "distillation_loss": 0.10803981125354767,
      "epoch": 7.12,
      "learning_rate": 4.9965654669972794e-05,
      "loss": 0.1114,
      "step": 7499,
      "task_loss": 0.14213261008262634
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998676054451607,
      "compression_loss": 0.0,
      "distillation_loss": 0.018904566764831543,
      "epoch": 7.12,
      "learning_rate": 4.99650940696784e-05,
      "loss": 0.0179,
      "step": 7500,
      "task_loss": 0.009208250790834427
    },
    {
      "epoch": 7.12,
      "eval_accuracy": 0.9036697247706422,
      "eval_loss": 0.4249095320701599,
      "eval_runtime": 18.0437,
      "eval_samples_per_second": 48.327,
      "eval_steps_per_second": 6.041,
      "step": 7500
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998680343688704,
      "compression_loss": 0.0,
      "distillation_loss": 0.1428883969783783,
      "epoch": 7.12,
      "learning_rate": 4.996452893435442e-05,
      "loss": 0.1441,
      "step": 7501,
      "task_loss": 0.1551024615764618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799868462365178,
      "compression_loss": 0.0,
      "distillation_loss": 0.04754265025258064,
      "epoch": 7.12,
      "learning_rate": 4.9963959264103544e-05,
      "loss": 0.0454,
      "step": 7502,
      "task_loss": 0.026247184723615646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998688894350868,
      "compression_loss": 0.0,
      "distillation_loss": 0.04800819978117943,
      "epoch": 7.13,
      "learning_rate": 4.996338505902924e-05,
      "loss": 0.0622,
      "step": 7503,
      "task_loss": 0.18986263871192932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998693155796008,
      "compression_loss": 0.0,
      "distillation_loss": 0.08775762468576431,
      "epoch": 7.13,
      "learning_rate": 4.996280631923581e-05,
      "loss": 0.0996,
      "step": 7504,
      "task_loss": 0.20661881566047668
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998697407997237,
      "compression_loss": 0.0,
      "distillation_loss": 0.05795777961611748,
      "epoch": 7.13,
      "learning_rate": 4.9962223044828396e-05,
      "loss": 0.0575,
      "step": 7505,
      "task_loss": 0.05344875901937485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998701650964589,
      "compression_loss": 0.0,
      "distillation_loss": 0.07826469838619232,
      "epoch": 7.13,
      "learning_rate": 4.9961635235912935e-05,
      "loss": 0.0739,
      "step": 7506,
      "task_loss": 0.03454684466123581
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998705884708104,
      "compression_loss": 0.0,
      "distillation_loss": 0.016553938388824463,
      "epoch": 7.13,
      "learning_rate": 4.9961042892596225e-05,
      "loss": 0.0304,
      "step": 7507,
      "task_loss": 0.1547098606824875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998710109237817,
      "compression_loss": 0.0,
      "distillation_loss": 0.024404142051935196,
      "epoch": 7.13,
      "learning_rate": 4.996044601498586e-05,
      "loss": 0.0231,
      "step": 7508,
      "task_loss": 0.010925725102424622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998714324563765,
      "compression_loss": 0.0,
      "distillation_loss": 0.12660256028175354,
      "epoch": 7.13,
      "learning_rate": 4.995984460319026e-05,
      "loss": 0.1258,
      "step": 7509,
      "task_loss": 0.11870937794446945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998718530695985,
      "compression_loss": 0.0,
      "distillation_loss": 0.029197368770837784,
      "epoch": 7.13,
      "learning_rate": 4.995923865731869e-05,
      "loss": 0.0291,
      "step": 7510,
      "task_loss": 0.028612298890948296
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998722727644514,
      "compression_loss": 0.0,
      "distillation_loss": 0.01047599595040083,
      "epoch": 7.13,
      "learning_rate": 4.9958628177481195e-05,
      "loss": 0.0097,
      "step": 7511,
      "task_loss": 0.002894926816225052
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998726915419389,
      "compression_loss": 0.0,
      "distillation_loss": 0.13094131648540497,
      "epoch": 7.13,
      "learning_rate": 4.99580131637887e-05,
      "loss": 0.1189,
      "step": 7512,
      "task_loss": 0.011026578024029732
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998731094030646,
      "compression_loss": 0.0,
      "distillation_loss": 0.054517075419425964,
      "epoch": 7.13,
      "learning_rate": 4.995739361635292e-05,
      "loss": 0.0499,
      "step": 7513,
      "task_loss": 0.008429093286395073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998735263488322,
      "compression_loss": 0.0,
      "distillation_loss": 0.057775288820266724,
      "epoch": 7.14,
      "learning_rate": 4.9956769535286385e-05,
      "loss": 0.0621,
      "step": 7514,
      "task_loss": 0.1013016402721405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998739423802455,
      "compression_loss": 0.0,
      "distillation_loss": 0.13451939821243286,
      "epoch": 7.14,
      "learning_rate": 4.9956140920702476e-05,
      "loss": 0.1393,
      "step": 7515,
      "task_loss": 0.1819526106119156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799874357498308,
      "compression_loss": 0.0,
      "distillation_loss": 0.039106979966163635,
      "epoch": 7.14,
      "learning_rate": 4.995550777271538e-05,
      "loss": 0.0363,
      "step": 7516,
      "task_loss": 0.010946827009320259
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998747717040235,
      "compression_loss": 0.0,
      "distillation_loss": 0.03019588068127632,
      "epoch": 7.14,
      "learning_rate": 4.995487009144011e-05,
      "loss": 0.0277,
      "step": 7517,
      "task_loss": 0.005600154399871826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998751849983957,
      "compression_loss": 0.0,
      "distillation_loss": 0.1031942293047905,
      "epoch": 7.14,
      "learning_rate": 4.99542278769925e-05,
      "loss": 0.1217,
      "step": 7518,
      "task_loss": 0.28847843408584595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998755973824282,
      "compression_loss": 0.0,
      "distillation_loss": 0.03711920231580734,
      "epoch": 7.14,
      "learning_rate": 4.995358112948921e-05,
      "loss": 0.0388,
      "step": 7519,
      "task_loss": 0.053622905164957047
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998760088571247,
      "compression_loss": 0.0,
      "distillation_loss": 0.036818671971559525,
      "epoch": 7.14,
      "learning_rate": 4.9952929849047734e-05,
      "loss": 0.034,
      "step": 7520,
      "task_loss": 0.008642930537462234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998764194234891,
      "compression_loss": 0.0,
      "distillation_loss": 0.040213510394096375,
      "epoch": 7.14,
      "learning_rate": 4.9952274035786385e-05,
      "loss": 0.0417,
      "step": 7521,
      "task_loss": 0.054693013429641724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998768290825247,
      "compression_loss": 0.0,
      "distillation_loss": 0.025412341579794884,
      "epoch": 7.14,
      "learning_rate": 4.9951613689824276e-05,
      "loss": 0.0302,
      "step": 7522,
      "task_loss": 0.07287220656871796
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998772378352353,
      "compression_loss": 0.0,
      "distillation_loss": 0.052676744759082794,
      "epoch": 7.14,
      "learning_rate": 4.995094881128138e-05,
      "loss": 0.0828,
      "step": 7523,
      "task_loss": 0.3543989956378937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998776456826249,
      "compression_loss": 0.0,
      "distillation_loss": 0.04211168363690376,
      "epoch": 7.15,
      "learning_rate": 4.995027940027846e-05,
      "loss": 0.0418,
      "step": 7524,
      "task_loss": 0.038644563406705856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998780526256968,
      "compression_loss": 0.0,
      "distillation_loss": 0.0226961188018322,
      "epoch": 7.15,
      "learning_rate": 4.9949605456937135e-05,
      "loss": 0.0388,
      "step": 7525,
      "task_loss": 0.18341752886772156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998784586654548,
      "compression_loss": 0.0,
      "distillation_loss": 0.05215848237276077,
      "epoch": 7.15,
      "learning_rate": 4.994892698137981e-05,
      "loss": 0.0506,
      "step": 7526,
      "task_loss": 0.03646836429834366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998788638029026,
      "compression_loss": 0.0,
      "distillation_loss": 0.022788241505622864,
      "epoch": 7.15,
      "learning_rate": 4.9948243973729745e-05,
      "loss": 0.029,
      "step": 7527,
      "task_loss": 0.08468222618103027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998792680390439,
      "compression_loss": 0.0,
      "distillation_loss": 0.08569545298814774,
      "epoch": 7.15,
      "learning_rate": 4.994755643411101e-05,
      "loss": 0.0933,
      "step": 7528,
      "task_loss": 0.16126233339309692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998796713748824,
      "compression_loss": 0.0,
      "distillation_loss": 0.019816741347312927,
      "epoch": 7.15,
      "learning_rate": 4.9946864362648506e-05,
      "loss": 0.0305,
      "step": 7529,
      "task_loss": 0.1267254650592804
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998800738114217,
      "compression_loss": 0.0,
      "distillation_loss": 0.17498815059661865,
      "epoch": 7.15,
      "learning_rate": 4.994616775946794e-05,
      "loss": 0.1664,
      "step": 7530,
      "task_loss": 0.08893725275993347
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998804753496654,
      "compression_loss": 0.0,
      "distillation_loss": 0.022720731794834137,
      "epoch": 7.15,
      "learning_rate": 4.994546662469586e-05,
      "loss": 0.0207,
      "step": 7531,
      "task_loss": 0.002466036006808281
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998808759906175,
      "compression_loss": 0.0,
      "distillation_loss": 0.045352548360824585,
      "epoch": 7.15,
      "learning_rate": 4.9944760958459624e-05,
      "loss": 0.0428,
      "step": 7532,
      "task_loss": 0.020112009719014168
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998812757352814,
      "compression_loss": 0.0,
      "distillation_loss": 0.059900932013988495,
      "epoch": 7.15,
      "learning_rate": 4.994405076088743e-05,
      "loss": 0.063,
      "step": 7533,
      "task_loss": 0.0908510833978653
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799881674584661,
      "compression_loss": 0.0,
      "distillation_loss": 0.06974376738071442,
      "epoch": 7.15,
      "learning_rate": 4.994333603210829e-05,
      "loss": 0.0684,
      "step": 7534,
      "task_loss": 0.05587990581989288
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998820725397597,
      "compression_loss": 0.0,
      "distillation_loss": 0.058896493166685104,
      "epoch": 7.16,
      "learning_rate": 4.9942616772252016e-05,
      "loss": 0.0566,
      "step": 7535,
      "task_loss": 0.0360528826713562
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998824696015814,
      "compression_loss": 0.0,
      "distillation_loss": 0.047747716307640076,
      "epoch": 7.16,
      "learning_rate": 4.994189298144929e-05,
      "loss": 0.0435,
      "step": 7536,
      "task_loss": 0.005159799009561539
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998828657711297,
      "compression_loss": 0.0,
      "distillation_loss": 0.019652074202895164,
      "epoch": 7.16,
      "learning_rate": 4.994116465983158e-05,
      "loss": 0.0345,
      "step": 7537,
      "task_loss": 0.16844874620437622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998832610494083,
      "compression_loss": 0.0,
      "distillation_loss": 0.17571541666984558,
      "epoch": 7.16,
      "learning_rate": 4.99404318075312e-05,
      "loss": 0.1646,
      "step": 7538,
      "task_loss": 0.06422274559736252
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998836554374209,
      "compression_loss": 0.0,
      "distillation_loss": 0.1369340866804123,
      "epoch": 7.16,
      "learning_rate": 4.993969442468125e-05,
      "loss": 0.1452,
      "step": 7539,
      "task_loss": 0.21913862228393555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998840489361712,
      "compression_loss": 0.0,
      "distillation_loss": 0.05932403728365898,
      "epoch": 7.16,
      "learning_rate": 4.993895251141571e-05,
      "loss": 0.0689,
      "step": 7540,
      "task_loss": 0.15557356178760529
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998844415466628,
      "compression_loss": 0.0,
      "distillation_loss": 0.04140976071357727,
      "epoch": 7.16,
      "learning_rate": 4.9938206067869334e-05,
      "loss": 0.0454,
      "step": 7541,
      "task_loss": 0.0813070684671402
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998848332698995,
      "compression_loss": 0.0,
      "distillation_loss": 0.036964789032936096,
      "epoch": 7.16,
      "learning_rate": 4.993745509417772e-05,
      "loss": 0.0396,
      "step": 7542,
      "task_loss": 0.06300060451030731
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998852241068848,
      "compression_loss": 0.0,
      "distillation_loss": 0.14604540169239044,
      "epoch": 7.16,
      "learning_rate": 4.9936699590477296e-05,
      "loss": 0.1421,
      "step": 7543,
      "task_loss": 0.10691849142313004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998856140586226,
      "compression_loss": 0.0,
      "distillation_loss": 0.10992075502872467,
      "epoch": 7.16,
      "learning_rate": 4.9935939556905295e-05,
      "loss": 0.1188,
      "step": 7544,
      "task_loss": 0.19906115531921387
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998860031261164,
      "compression_loss": 0.0,
      "distillation_loss": 0.014540485106408596,
      "epoch": 7.17,
      "learning_rate": 4.993517499359978e-05,
      "loss": 0.0135,
      "step": 7545,
      "task_loss": 0.003651769831776619
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79988639131037,
      "compression_loss": 0.0,
      "distillation_loss": 0.06785355508327484,
      "epoch": 7.17,
      "learning_rate": 4.993440590069963e-05,
      "loss": 0.0756,
      "step": 7546,
      "task_loss": 0.1453818380832672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799886778612387,
      "compression_loss": 0.0,
      "distillation_loss": 0.08633054792881012,
      "epoch": 7.17,
      "learning_rate": 4.993363227834457e-05,
      "loss": 0.0958,
      "step": 7547,
      "task_loss": 0.18063369393348694
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998871650331711,
      "compression_loss": 0.0,
      "distillation_loss": 0.13422691822052002,
      "epoch": 7.17,
      "learning_rate": 4.9932854126675124e-05,
      "loss": 0.1375,
      "step": 7548,
      "task_loss": 0.16670459508895874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998875505737261,
      "compression_loss": 0.0,
      "distillation_loss": 0.0358545184135437,
      "epoch": 7.17,
      "learning_rate": 4.993207144583264e-05,
      "loss": 0.0333,
      "step": 7549,
      "task_loss": 0.009983271360397339
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998879352350555,
      "compression_loss": 0.0,
      "distillation_loss": 0.03655195236206055,
      "epoch": 7.17,
      "learning_rate": 4.993128423595931e-05,
      "loss": 0.0333,
      "step": 7550,
      "task_loss": 0.0044655874371528625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799888319018163,
      "compression_loss": 0.0,
      "distillation_loss": 0.04761989042162895,
      "epoch": 7.17,
      "learning_rate": 4.9930492497198125e-05,
      "loss": 0.0541,
      "step": 7551,
      "task_loss": 0.11243405193090439
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998887019240526,
      "compression_loss": 0.0,
      "distillation_loss": 0.04824792221188545,
      "epoch": 7.17,
      "learning_rate": 4.992969622969292e-05,
      "loss": 0.0455,
      "step": 7552,
      "task_loss": 0.020604906603693962
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998890839537275,
      "compression_loss": 0.0,
      "distillation_loss": 0.02487030252814293,
      "epoch": 7.17,
      "learning_rate": 4.992889543358832e-05,
      "loss": 0.0281,
      "step": 7553,
      "task_loss": 0.05736660212278366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998894651081917,
      "compression_loss": 0.0,
      "distillation_loss": 0.040249839425086975,
      "epoch": 7.17,
      "learning_rate": 4.9928090109029817e-05,
      "loss": 0.0528,
      "step": 7554,
      "task_loss": 0.16572295129299164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998898453884489,
      "compression_loss": 0.0,
      "distillation_loss": 0.08049503713846207,
      "epoch": 7.17,
      "learning_rate": 4.9927280256163686e-05,
      "loss": 0.0842,
      "step": 7555,
      "task_loss": 0.11797265708446503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998902247955025,
      "compression_loss": 0.0,
      "distillation_loss": 0.0269448421895504,
      "epoch": 7.18,
      "learning_rate": 4.992646587513705e-05,
      "loss": 0.0313,
      "step": 7556,
      "task_loss": 0.07026375085115433
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998906033303564,
      "compression_loss": 0.0,
      "distillation_loss": 0.03366638720035553,
      "epoch": 7.18,
      "learning_rate": 4.9925646966097835e-05,
      "loss": 0.0313,
      "step": 7557,
      "task_loss": 0.010441986843943596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998909809940143,
      "compression_loss": 0.0,
      "distillation_loss": 0.015848444774746895,
      "epoch": 7.18,
      "learning_rate": 4.99248235291948e-05,
      "loss": 0.0309,
      "step": 7558,
      "task_loss": 0.16670146584510803
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998913577874799,
      "compression_loss": 0.0,
      "distillation_loss": 0.05733553692698479,
      "epoch": 7.18,
      "learning_rate": 4.9923995564577544e-05,
      "loss": 0.0547,
      "step": 7559,
      "task_loss": 0.03086872212588787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998917337117567,
      "compression_loss": 0.0,
      "distillation_loss": 0.03503969684243202,
      "epoch": 7.18,
      "learning_rate": 4.992316307239645e-05,
      "loss": 0.0492,
      "step": 7560,
      "task_loss": 0.17629651725292206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998921087678484,
      "compression_loss": 0.0,
      "distillation_loss": 0.07000732421875,
      "epoch": 7.18,
      "learning_rate": 4.992232605280276e-05,
      "loss": 0.078,
      "step": 7561,
      "task_loss": 0.14994311332702637
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799892482956759,
      "compression_loss": 0.0,
      "distillation_loss": 0.042889855802059174,
      "epoch": 7.18,
      "learning_rate": 4.992148450594851e-05,
      "loss": 0.039,
      "step": 7562,
      "task_loss": 0.0036024488508701324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998928562794918,
      "compression_loss": 0.0,
      "distillation_loss": 0.012830915860831738,
      "epoch": 7.18,
      "learning_rate": 4.9920638431986574e-05,
      "loss": 0.0119,
      "step": 7563,
      "task_loss": 0.0039144158363342285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998932287370506,
      "compression_loss": 0.0,
      "distillation_loss": 0.017170187085866928,
      "epoch": 7.18,
      "learning_rate": 4.991978783107065e-05,
      "loss": 0.0199,
      "step": 7564,
      "task_loss": 0.04400904104113579
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998936003304392,
      "compression_loss": 0.0,
      "distillation_loss": 0.019753381609916687,
      "epoch": 7.18,
      "learning_rate": 4.9918932703355256e-05,
      "loss": 0.0186,
      "step": 7565,
      "task_loss": 0.008108856156468391
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998939710606612,
      "compression_loss": 0.0,
      "distillation_loss": 0.11725122481584549,
      "epoch": 7.19,
      "learning_rate": 4.991807304899572e-05,
      "loss": 0.1261,
      "step": 7566,
      "task_loss": 0.20566239953041077
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998943409287202,
      "compression_loss": 0.0,
      "distillation_loss": 0.06889253109693527,
      "epoch": 7.19,
      "learning_rate": 4.991720886814821e-05,
      "loss": 0.0706,
      "step": 7567,
      "task_loss": 0.08627396821975708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998947099356201,
      "compression_loss": 0.0,
      "distillation_loss": 0.13849744200706482,
      "epoch": 7.19,
      "learning_rate": 4.99163401609697e-05,
      "loss": 0.1404,
      "step": 7568,
      "task_loss": 0.15727216005325317
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998950780823644,
      "compression_loss": 0.0,
      "distillation_loss": 0.0490923747420311,
      "epoch": 7.19,
      "learning_rate": 4.991546692761801e-05,
      "loss": 0.047,
      "step": 7569,
      "task_loss": 0.028280524536967278
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998954453699567,
      "compression_loss": 0.0,
      "distillation_loss": 0.0508795864880085,
      "epoch": 7.19,
      "learning_rate": 4.991458916825176e-05,
      "loss": 0.0486,
      "step": 7570,
      "task_loss": 0.027954697608947754
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799895811799401,
      "compression_loss": 0.0,
      "distillation_loss": 0.023934636265039444,
      "epoch": 7.19,
      "learning_rate": 4.991370688303039e-05,
      "loss": 0.0246,
      "step": 7571,
      "task_loss": 0.030092671513557434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998961773717007,
      "compression_loss": 0.0,
      "distillation_loss": 0.016041746363043785,
      "epoch": 7.19,
      "learning_rate": 4.9912820072114185e-05,
      "loss": 0.0149,
      "step": 7572,
      "task_loss": 0.0041885413229465485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998965420878595,
      "compression_loss": 0.0,
      "distillation_loss": 0.051787007600069046,
      "epoch": 7.19,
      "learning_rate": 4.9911928735664224e-05,
      "loss": 0.0531,
      "step": 7573,
      "task_loss": 0.0652824267745018
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998969059488813,
      "compression_loss": 0.0,
      "distillation_loss": 0.11679676175117493,
      "epoch": 7.19,
      "learning_rate": 4.991103287384244e-05,
      "loss": 0.1229,
      "step": 7574,
      "task_loss": 0.1777617633342743
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998972689557696,
      "compression_loss": 0.0,
      "distillation_loss": 0.02287183701992035,
      "epoch": 7.19,
      "learning_rate": 4.9910132486811555e-05,
      "loss": 0.0216,
      "step": 7575,
      "task_loss": 0.009705502539873123
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799897631109528,
      "compression_loss": 0.0,
      "distillation_loss": 0.07535935938358307,
      "epoch": 7.19,
      "learning_rate": 4.990922757473514e-05,
      "loss": 0.0712,
      "step": 7576,
      "task_loss": 0.034015409648418427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998979924111604,
      "compression_loss": 0.0,
      "distillation_loss": 0.03653284162282944,
      "epoch": 7.2,
      "learning_rate": 4.990831813777757e-05,
      "loss": 0.0336,
      "step": 7577,
      "task_loss": 0.006952434778213501
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998983528616704,
      "compression_loss": 0.0,
      "distillation_loss": 0.06351464986801147,
      "epoch": 7.2,
      "learning_rate": 4.990740417610406e-05,
      "loss": 0.0599,
      "step": 7578,
      "task_loss": 0.02752096578478813
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998987124620617,
      "compression_loss": 0.0,
      "distillation_loss": 0.019939813762903214,
      "epoch": 7.2,
      "learning_rate": 4.9906485689880613e-05,
      "loss": 0.0185,
      "step": 7579,
      "task_loss": 0.005348971113562584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998990712133379,
      "compression_loss": 0.0,
      "distillation_loss": 0.014425119385123253,
      "epoch": 7.2,
      "learning_rate": 4.9905562679274096e-05,
      "loss": 0.0141,
      "step": 7580,
      "task_loss": 0.011267106980085373
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7998994291165028,
      "compression_loss": 0.0,
      "distillation_loss": 0.1040351465344429,
      "epoch": 7.2,
      "learning_rate": 4.9904635144452164e-05,
      "loss": 0.1014,
      "step": 7581,
      "task_loss": 0.07765233516693115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79989978617256,
      "compression_loss": 0.0,
      "distillation_loss": 0.049760229885578156,
      "epoch": 7.2,
      "learning_rate": 4.990370308558332e-05,
      "loss": 0.0527,
      "step": 7582,
      "task_loss": 0.07866282761096954
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999001423825131,
      "compression_loss": 0.0,
      "distillation_loss": 0.014256338588893414,
      "epoch": 7.2,
      "learning_rate": 4.9902766502836874e-05,
      "loss": 0.0235,
      "step": 7583,
      "task_loss": 0.10717885941267014
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999004977473658,
      "compression_loss": 0.0,
      "distillation_loss": 0.084557443857193,
      "epoch": 7.2,
      "learning_rate": 4.9901825396382965e-05,
      "loss": 0.0891,
      "step": 7584,
      "task_loss": 0.13022615015506744
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799900852268122,
      "compression_loss": 0.0,
      "distillation_loss": 0.1039082333445549,
      "epoch": 7.2,
      "learning_rate": 4.990087976639254e-05,
      "loss": 0.1037,
      "step": 7585,
      "task_loss": 0.10136047005653381
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999012059457853,
      "compression_loss": 0.0,
      "distillation_loss": 0.08127206563949585,
      "epoch": 7.2,
      "learning_rate": 4.989992961303738e-05,
      "loss": 0.0907,
      "step": 7586,
      "task_loss": 0.17568367719650269
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999015587813593,
      "compression_loss": 0.0,
      "distillation_loss": 0.012685288675129414,
      "epoch": 7.21,
      "learning_rate": 4.989897493649008e-05,
      "loss": 0.0195,
      "step": 7587,
      "task_loss": 0.08097560703754425
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999019107758476,
      "compression_loss": 0.0,
      "distillation_loss": 0.03525643050670624,
      "epoch": 7.21,
      "learning_rate": 4.989801573692408e-05,
      "loss": 0.0407,
      "step": 7588,
      "task_loss": 0.08939573168754578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799902261930254,
      "compression_loss": 0.0,
      "distillation_loss": 0.04309917986392975,
      "epoch": 7.21,
      "learning_rate": 4.989705201451361e-05,
      "loss": 0.0391,
      "step": 7589,
      "task_loss": 0.00266990065574646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999026122455822,
      "compression_loss": 0.0,
      "distillation_loss": 0.018025081604719162,
      "epoch": 7.21,
      "learning_rate": 4.989608376943373e-05,
      "loss": 0.0172,
      "step": 7590,
      "task_loss": 0.009299803525209427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999029617228359,
      "compression_loss": 0.0,
      "distillation_loss": 0.011557912454009056,
      "epoch": 7.21,
      "learning_rate": 4.9895111001860335e-05,
      "loss": 0.0108,
      "step": 7591,
      "task_loss": 0.004404161125421524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999033103630186,
      "compression_loss": 0.0,
      "distillation_loss": 0.021339278668165207,
      "epoch": 7.21,
      "learning_rate": 4.989413371197013e-05,
      "loss": 0.0201,
      "step": 7592,
      "task_loss": 0.008754091337323189
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999036581671343,
      "compression_loss": 0.0,
      "distillation_loss": 0.08135688304901123,
      "epoch": 7.21,
      "learning_rate": 4.989315189994065e-05,
      "loss": 0.087,
      "step": 7593,
      "task_loss": 0.13778801262378693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999040051361864,
      "compression_loss": 0.0,
      "distillation_loss": 0.020718682557344437,
      "epoch": 7.21,
      "learning_rate": 4.9892165565950235e-05,
      "loss": 0.0198,
      "step": 7594,
      "task_loss": 0.011542653664946556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999043512711788,
      "compression_loss": 0.0,
      "distillation_loss": 0.029304485768079758,
      "epoch": 7.21,
      "learning_rate": 4.9891174710178054e-05,
      "loss": 0.0359,
      "step": 7595,
      "task_loss": 0.09544922411441803
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999046965731149,
      "compression_loss": 0.0,
      "distillation_loss": 0.012081381864845753,
      "epoch": 7.21,
      "learning_rate": 4.9890179332804125e-05,
      "loss": 0.0112,
      "step": 7596,
      "task_loss": 0.003652777522802353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999050410429985,
      "compression_loss": 0.0,
      "distillation_loss": 0.032954856753349304,
      "epoch": 7.21,
      "learning_rate": 4.988917943400924e-05,
      "loss": 0.032,
      "step": 7597,
      "task_loss": 0.023814279586076736
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999053846818335,
      "compression_loss": 0.0,
      "distillation_loss": 0.019010456278920174,
      "epoch": 7.22,
      "learning_rate": 4.988817501397505e-05,
      "loss": 0.0209,
      "step": 7598,
      "task_loss": 0.03756894916296005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999057274906234,
      "compression_loss": 0.0,
      "distillation_loss": 0.05136597529053688,
      "epoch": 7.22,
      "learning_rate": 4.9887166072884e-05,
      "loss": 0.0477,
      "step": 7599,
      "task_loss": 0.014553211629390717
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999060694703718,
      "compression_loss": 0.0,
      "distillation_loss": 0.02315210923552513,
      "epoch": 7.22,
      "learning_rate": 4.988615261091938e-05,
      "loss": 0.0213,
      "step": 7600,
      "task_loss": 0.004539225250482559
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999064106220825,
      "compression_loss": 0.0,
      "distillation_loss": 0.05613788589835167,
      "epoch": 7.22,
      "learning_rate": 4.9885134628265276e-05,
      "loss": 0.0686,
      "step": 7601,
      "task_loss": 0.1807435154914856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999067509467591,
      "compression_loss": 0.0,
      "distillation_loss": 0.017415829002857208,
      "epoch": 7.22,
      "learning_rate": 4.988411212510663e-05,
      "loss": 0.0263,
      "step": 7602,
      "task_loss": 0.10614414513111115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999070904454054,
      "compression_loss": 0.0,
      "distillation_loss": 0.02971852757036686,
      "epoch": 7.22,
      "learning_rate": 4.988308510162917e-05,
      "loss": 0.0275,
      "step": 7603,
      "task_loss": 0.007290884852409363
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799907429119025,
      "compression_loss": 0.0,
      "distillation_loss": 0.09431088715791702,
      "epoch": 7.22,
      "learning_rate": 4.988205355801945e-05,
      "loss": 0.0992,
      "step": 7604,
      "task_loss": 0.14289163053035736
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999077669686216,
      "compression_loss": 0.0,
      "distillation_loss": 0.11817564815282822,
      "epoch": 7.22,
      "learning_rate": 4.988101749446488e-05,
      "loss": 0.1131,
      "step": 7605,
      "task_loss": 0.06737690418958664
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799908103995199,
      "compression_loss": 0.0,
      "distillation_loss": 0.05987474322319031,
      "epoch": 7.22,
      "learning_rate": 4.987997691115366e-05,
      "loss": 0.0599,
      "step": 7606,
      "task_loss": 0.060150373727083206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999084401997606,
      "compression_loss": 0.0,
      "distillation_loss": 0.1141742616891861,
      "epoch": 7.22,
      "learning_rate": 4.98789318082748e-05,
      "loss": 0.1098,
      "step": 7607,
      "task_loss": 0.07077798992395401
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999087755833103,
      "compression_loss": 0.0,
      "distillation_loss": 0.024525757879018784,
      "epoch": 7.23,
      "learning_rate": 4.987788218601816e-05,
      "loss": 0.0302,
      "step": 7608,
      "task_loss": 0.08100198209285736
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999091101468517,
      "compression_loss": 0.0,
      "distillation_loss": 0.040195684880018234,
      "epoch": 7.23,
      "learning_rate": 4.987682804457441e-05,
      "loss": 0.0451,
      "step": 7609,
      "task_loss": 0.08934106677770615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999094438913886,
      "compression_loss": 0.0,
      "distillation_loss": 0.044671397656202316,
      "epoch": 7.23,
      "learning_rate": 4.987576938413504e-05,
      "loss": 0.0412,
      "step": 7610,
      "task_loss": 0.009734295308589935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999097768179245,
      "compression_loss": 0.0,
      "distillation_loss": 0.017387911677360535,
      "epoch": 7.23,
      "learning_rate": 4.987470620489235e-05,
      "loss": 0.0164,
      "step": 7611,
      "task_loss": 0.007594836875796318
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999101089274633,
      "compression_loss": 0.0,
      "distillation_loss": 0.04098183289170265,
      "epoch": 7.23,
      "learning_rate": 4.9873638507039486e-05,
      "loss": 0.0408,
      "step": 7612,
      "task_loss": 0.039046209305524826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999104402210084,
      "compression_loss": 0.0,
      "distillation_loss": 0.0328001007437706,
      "epoch": 7.23,
      "learning_rate": 4.987256629077039e-05,
      "loss": 0.0379,
      "step": 7613,
      "task_loss": 0.08381561934947968
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999107706995638,
      "compression_loss": 0.0,
      "distillation_loss": 0.06902207434177399,
      "epoch": 7.23,
      "learning_rate": 4.987148955627985e-05,
      "loss": 0.0674,
      "step": 7614,
      "task_loss": 0.05302996188402176
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999111003641329,
      "compression_loss": 0.0,
      "distillation_loss": 0.015985846519470215,
      "epoch": 7.23,
      "learning_rate": 4.987040830376344e-05,
      "loss": 0.0202,
      "step": 7615,
      "task_loss": 0.05846152827143669
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999114292157196,
      "compression_loss": 0.0,
      "distillation_loss": 0.04634019732475281,
      "epoch": 7.23,
      "learning_rate": 4.9869322533417596e-05,
      "loss": 0.0425,
      "step": 7616,
      "task_loss": 0.008089037612080574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999117572553275,
      "compression_loss": 0.0,
      "distillation_loss": 0.0537746362388134,
      "epoch": 7.23,
      "learning_rate": 4.9868232245439525e-05,
      "loss": 0.0524,
      "step": 7617,
      "task_loss": 0.039632245898246765
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999120844839601,
      "compression_loss": 0.0,
      "distillation_loss": 0.05178552865982056,
      "epoch": 7.23,
      "learning_rate": 4.986713744002731e-05,
      "loss": 0.0538,
      "step": 7618,
      "task_loss": 0.07191041857004166
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999124109026214,
      "compression_loss": 0.0,
      "distillation_loss": 0.025834370404481888,
      "epoch": 7.24,
      "learning_rate": 4.9866038117379824e-05,
      "loss": 0.0236,
      "step": 7619,
      "task_loss": 0.003722256049513817
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999127365123149,
      "compression_loss": 0.0,
      "distillation_loss": 0.19725723564624786,
      "epoch": 7.24,
      "learning_rate": 4.986493427769675e-05,
      "loss": 0.1893,
      "step": 7620,
      "task_loss": 0.11761374771595001
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999130613140443,
      "compression_loss": 0.0,
      "distillation_loss": 0.05075707659125328,
      "epoch": 7.24,
      "learning_rate": 4.986382592117861e-05,
      "loss": 0.0515,
      "step": 7621,
      "task_loss": 0.05864773318171501
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999133853088133,
      "compression_loss": 0.0,
      "distillation_loss": 0.021583644673228264,
      "epoch": 7.24,
      "learning_rate": 4.986271304802675e-05,
      "loss": 0.0223,
      "step": 7622,
      "task_loss": 0.028752833604812622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999137084976256,
      "compression_loss": 0.0,
      "distillation_loss": 0.12262220680713654,
      "epoch": 7.24,
      "learning_rate": 4.986159565844333e-05,
      "loss": 0.1209,
      "step": 7623,
      "task_loss": 0.10493774712085724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999140308814848,
      "compression_loss": 0.0,
      "distillation_loss": 0.054139602929353714,
      "epoch": 7.24,
      "learning_rate": 4.986047375263131e-05,
      "loss": 0.0749,
      "step": 7624,
      "task_loss": 0.2614555060863495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999143524613948,
      "compression_loss": 0.0,
      "distillation_loss": 0.011302494443953037,
      "epoch": 7.24,
      "learning_rate": 4.9859347330794515e-05,
      "loss": 0.0164,
      "step": 7625,
      "task_loss": 0.0621810182929039
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799914673238359,
      "compression_loss": 0.0,
      "distillation_loss": 0.019549107179045677,
      "epoch": 7.24,
      "learning_rate": 4.985821639313755e-05,
      "loss": 0.0179,
      "step": 7626,
      "task_loss": 0.003164045512676239
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999149932133811,
      "compression_loss": 0.0,
      "distillation_loss": 0.07022275030612946,
      "epoch": 7.24,
      "learning_rate": 4.985708093986586e-05,
      "loss": 0.0732,
      "step": 7627,
      "task_loss": 0.09962170571088791
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799915312387465,
      "compression_loss": 0.0,
      "distillation_loss": 0.042909909039735794,
      "epoch": 7.24,
      "learning_rate": 4.98559409711857e-05,
      "loss": 0.0394,
      "step": 7628,
      "task_loss": 0.007330624386668205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999156307616143,
      "compression_loss": 0.0,
      "distillation_loss": 0.043095022439956665,
      "epoch": 7.25,
      "learning_rate": 4.985479648730416e-05,
      "loss": 0.0452,
      "step": 7629,
      "task_loss": 0.06463564932346344
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999159483368327,
      "compression_loss": 0.0,
      "distillation_loss": 0.05689408257603645,
      "epoch": 7.25,
      "learning_rate": 4.985364748842914e-05,
      "loss": 0.0657,
      "step": 7630,
      "task_loss": 0.14448405802249908
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999162651141237,
      "compression_loss": 0.0,
      "distillation_loss": 0.09874869883060455,
      "epoch": 7.25,
      "learning_rate": 4.985249397476934e-05,
      "loss": 0.103,
      "step": 7631,
      "task_loss": 0.14085200428962708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999165810944913,
      "compression_loss": 0.0,
      "distillation_loss": 0.10595101863145828,
      "epoch": 7.25,
      "learning_rate": 4.985133594653434e-05,
      "loss": 0.1023,
      "step": 7632,
      "task_loss": 0.06989569962024689
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999168962789388,
      "compression_loss": 0.0,
      "distillation_loss": 0.11988788843154907,
      "epoch": 7.25,
      "learning_rate": 4.9850173403934466e-05,
      "loss": 0.1167,
      "step": 7633,
      "task_loss": 0.08802229911088943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999172106684702,
      "compression_loss": 0.0,
      "distillation_loss": 0.026174481958150864,
      "epoch": 7.25,
      "learning_rate": 4.9849006347180915e-05,
      "loss": 0.0297,
      "step": 7634,
      "task_loss": 0.06151473894715309
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799917524264089,
      "compression_loss": 0.0,
      "distillation_loss": 0.02765459194779396,
      "epoch": 7.25,
      "learning_rate": 4.9847834776485694e-05,
      "loss": 0.0379,
      "step": 7635,
      "task_loss": 0.12964758276939392
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999178370667991,
      "compression_loss": 0.0,
      "distillation_loss": 0.02943515032529831,
      "epoch": 7.25,
      "learning_rate": 4.984665869206161e-05,
      "loss": 0.0346,
      "step": 7636,
      "task_loss": 0.08072115480899811
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999181490776038,
      "compression_loss": 0.0,
      "distillation_loss": 0.09545106440782547,
      "epoch": 7.25,
      "learning_rate": 4.984547809412231e-05,
      "loss": 0.0933,
      "step": 7637,
      "task_loss": 0.07358794659376144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999184602975071,
      "compression_loss": 0.0,
      "distillation_loss": 0.030496662482619286,
      "epoch": 7.25,
      "learning_rate": 4.984429298288227e-05,
      "loss": 0.0326,
      "step": 7638,
      "task_loss": 0.051075611263513565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999187707275127,
      "compression_loss": 0.0,
      "distillation_loss": 0.07517765462398529,
      "epoch": 7.25,
      "learning_rate": 4.984310335855674e-05,
      "loss": 0.0973,
      "step": 7639,
      "task_loss": 0.29597049951553345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799919080368624,
      "compression_loss": 0.0,
      "distillation_loss": 0.06372272223234177,
      "epoch": 7.26,
      "learning_rate": 4.9841909221361855e-05,
      "loss": 0.0631,
      "step": 7640,
      "task_loss": 0.05780046060681343
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799919389221845,
      "compression_loss": 0.0,
      "distillation_loss": 0.041298843920230865,
      "epoch": 7.26,
      "learning_rate": 4.9840710571514515e-05,
      "loss": 0.0489,
      "step": 7641,
      "task_loss": 0.1176719143986702
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999196972881792,
      "compression_loss": 0.0,
      "distillation_loss": 0.027051087468862534,
      "epoch": 7.26,
      "learning_rate": 4.9839507409232464e-05,
      "loss": 0.0249,
      "step": 7642,
      "task_loss": 0.005155773833394051
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999200045686303,
      "compression_loss": 0.0,
      "distillation_loss": 0.08592119812965393,
      "epoch": 7.26,
      "learning_rate": 4.983829973473426e-05,
      "loss": 0.0866,
      "step": 7643,
      "task_loss": 0.09253047406673431
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799920311064202,
      "compression_loss": 0.0,
      "distillation_loss": 0.04047606140375137,
      "epoch": 7.26,
      "learning_rate": 4.983708754823929e-05,
      "loss": 0.0374,
      "step": 7644,
      "task_loss": 0.009331891313195229
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799920616775898,
      "compression_loss": 0.0,
      "distillation_loss": 0.058635760098695755,
      "epoch": 7.26,
      "learning_rate": 4.983587084996776e-05,
      "loss": 0.0624,
      "step": 7645,
      "task_loss": 0.09588063508272171
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799920921704722,
      "compression_loss": 0.0,
      "distillation_loss": 0.20887014269828796,
      "epoch": 7.26,
      "learning_rate": 4.9834649640140664e-05,
      "loss": 0.1998,
      "step": 7646,
      "task_loss": 0.11826451122760773
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999212258516776,
      "compression_loss": 0.0,
      "distillation_loss": 0.0295681431889534,
      "epoch": 7.26,
      "learning_rate": 4.9833423918979864e-05,
      "loss": 0.0274,
      "step": 7647,
      "task_loss": 0.00763612799346447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999215292177686,
      "compression_loss": 0.0,
      "distillation_loss": 0.03998691588640213,
      "epoch": 7.26,
      "learning_rate": 4.983219368670801e-05,
      "loss": 0.0477,
      "step": 7648,
      "task_loss": 0.117280974984169
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999218318039986,
      "compression_loss": 0.0,
      "distillation_loss": 0.04406122863292694,
      "epoch": 7.26,
      "learning_rate": 4.983095894354858e-05,
      "loss": 0.0483,
      "step": 7649,
      "task_loss": 0.08681491017341614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999221336113712,
      "compression_loss": 0.0,
      "distillation_loss": 0.02541101723909378,
      "epoch": 7.26,
      "learning_rate": 4.9829719689725865e-05,
      "loss": 0.0278,
      "step": 7650,
      "task_loss": 0.04967557266354561
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999224346408902,
      "compression_loss": 0.0,
      "distillation_loss": 0.01671786792576313,
      "epoch": 7.27,
      "learning_rate": 4.982847592546499e-05,
      "loss": 0.0154,
      "step": 7651,
      "task_loss": 0.0038224849849939346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999227348935593,
      "compression_loss": 0.0,
      "distillation_loss": 0.03665540739893913,
      "epoch": 7.27,
      "learning_rate": 4.982722765099189e-05,
      "loss": 0.0429,
      "step": 7652,
      "task_loss": 0.09867965430021286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999230343703821,
      "compression_loss": 0.0,
      "distillation_loss": 0.044541992247104645,
      "epoch": 7.27,
      "learning_rate": 4.982597486653332e-05,
      "loss": 0.0519,
      "step": 7653,
      "task_loss": 0.11796452105045319
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999233330723624,
      "compression_loss": 0.0,
      "distillation_loss": 0.030439257621765137,
      "epoch": 7.27,
      "learning_rate": 4.982471757231685e-05,
      "loss": 0.0352,
      "step": 7654,
      "task_loss": 0.07802172005176544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999236310005038,
      "compression_loss": 0.0,
      "distillation_loss": 0.0167954470962286,
      "epoch": 7.27,
      "learning_rate": 4.982345576857087e-05,
      "loss": 0.0169,
      "step": 7655,
      "task_loss": 0.017974717542529106
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999239281558099,
      "compression_loss": 0.0,
      "distillation_loss": 0.07342584431171417,
      "epoch": 7.27,
      "learning_rate": 4.9822189455524604e-05,
      "loss": 0.0754,
      "step": 7656,
      "task_loss": 0.09314364194869995
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999242245392846,
      "compression_loss": 0.0,
      "distillation_loss": 0.1788443773984909,
      "epoch": 7.27,
      "learning_rate": 4.982091863340808e-05,
      "loss": 0.1738,
      "step": 7657,
      "task_loss": 0.12857407331466675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999245201519313,
      "compression_loss": 0.0,
      "distillation_loss": 0.12496727705001831,
      "epoch": 7.27,
      "learning_rate": 4.9819643302452146e-05,
      "loss": 0.1192,
      "step": 7658,
      "task_loss": 0.06772095710039139
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999248149947539,
      "compression_loss": 0.0,
      "distillation_loss": 0.03145987540483475,
      "epoch": 7.27,
      "learning_rate": 4.981836346288847e-05,
      "loss": 0.029,
      "step": 7659,
      "task_loss": 0.00681840255856514
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799925109068756,
      "compression_loss": 0.0,
      "distillation_loss": 0.010385725647211075,
      "epoch": 7.27,
      "learning_rate": 4.981707911494955e-05,
      "loss": 0.0098,
      "step": 7660,
      "task_loss": 0.004181254655122757
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999254023749414,
      "compression_loss": 0.0,
      "distillation_loss": 0.011662168428301811,
      "epoch": 7.28,
      "learning_rate": 4.981579025886868e-05,
      "loss": 0.0109,
      "step": 7661,
      "task_loss": 0.0035572052001953125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999256949143135,
      "compression_loss": 0.0,
      "distillation_loss": 0.02936125546693802,
      "epoch": 7.28,
      "learning_rate": 4.981449689488e-05,
      "loss": 0.0453,
      "step": 7662,
      "task_loss": 0.18874679505825043
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999259866878764,
      "compression_loss": 0.0,
      "distillation_loss": 0.07481378316879272,
      "epoch": 7.28,
      "learning_rate": 4.981319902321846e-05,
      "loss": 0.072,
      "step": 7663,
      "task_loss": 0.04644451290369034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999262776966334,
      "compression_loss": 0.0,
      "distillation_loss": 0.10028335452079773,
      "epoch": 7.28,
      "learning_rate": 4.981189664411981e-05,
      "loss": 0.1078,
      "step": 7664,
      "task_loss": 0.17516379058361053
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999265679415883,
      "compression_loss": 0.0,
      "distillation_loss": 0.02721228078007698,
      "epoch": 7.28,
      "learning_rate": 4.981058975782063e-05,
      "loss": 0.034,
      "step": 7665,
      "task_loss": 0.09542686492204666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999268574237448,
      "compression_loss": 0.0,
      "distillation_loss": 0.035428885370492935,
      "epoch": 7.28,
      "learning_rate": 4.9809278364558336e-05,
      "loss": 0.0378,
      "step": 7666,
      "task_loss": 0.05906771123409271
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999271461441066,
      "compression_loss": 0.0,
      "distillation_loss": 0.04637853801250458,
      "epoch": 7.28,
      "learning_rate": 4.980796246457115e-05,
      "loss": 0.0482,
      "step": 7667,
      "task_loss": 0.06469672918319702
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999274341036775,
      "compression_loss": 0.0,
      "distillation_loss": 0.060008976608514786,
      "epoch": 7.28,
      "learning_rate": 4.9806642058098105e-05,
      "loss": 0.0557,
      "step": 7668,
      "task_loss": 0.016873404383659363
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799927721303461,
      "compression_loss": 0.0,
      "distillation_loss": 0.02527848817408085,
      "epoch": 7.28,
      "learning_rate": 4.980531714537905e-05,
      "loss": 0.0239,
      "step": 7669,
      "task_loss": 0.011759728193283081
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999280077444608,
      "compression_loss": 0.0,
      "distillation_loss": 0.021292556077241898,
      "epoch": 7.28,
      "learning_rate": 4.980398772665468e-05,
      "loss": 0.0197,
      "step": 7670,
      "task_loss": 0.005023036152124405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999282934276807,
      "compression_loss": 0.0,
      "distillation_loss": 0.028033584356307983,
      "epoch": 7.28,
      "learning_rate": 4.980265380216649e-05,
      "loss": 0.0286,
      "step": 7671,
      "task_loss": 0.03369821235537529
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999285783541241,
      "compression_loss": 0.0,
      "distillation_loss": 0.028027424588799477,
      "epoch": 7.29,
      "learning_rate": 4.9801315372156775e-05,
      "loss": 0.0488,
      "step": 7672,
      "task_loss": 0.23589852452278137
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999288625247951,
      "compression_loss": 0.0,
      "distillation_loss": 0.048800766468048096,
      "epoch": 7.29,
      "learning_rate": 4.979997243686868e-05,
      "loss": 0.0462,
      "step": 7673,
      "task_loss": 0.022747982293367386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999291459406971,
      "compression_loss": 0.0,
      "distillation_loss": 0.040504857897758484,
      "epoch": 7.29,
      "learning_rate": 4.979862499654615e-05,
      "loss": 0.0495,
      "step": 7674,
      "task_loss": 0.13015000522136688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999294286028338,
      "compression_loss": 0.0,
      "distillation_loss": 0.017775043845176697,
      "epoch": 7.29,
      "learning_rate": 4.9797273051433966e-05,
      "loss": 0.0164,
      "step": 7675,
      "task_loss": 0.004127055406570435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799929710512209,
      "compression_loss": 0.0,
      "distillation_loss": 0.13806067407131195,
      "epoch": 7.29,
      "learning_rate": 4.97959166017777e-05,
      "loss": 0.1467,
      "step": 7676,
      "task_loss": 0.2246706783771515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999299916698263,
      "compression_loss": 0.0,
      "distillation_loss": 0.05745643004775047,
      "epoch": 7.29,
      "learning_rate": 4.979455564782377e-05,
      "loss": 0.0535,
      "step": 7677,
      "task_loss": 0.017686685547232628
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999302720766893,
      "compression_loss": 0.0,
      "distillation_loss": 0.07123734056949615,
      "epoch": 7.29,
      "learning_rate": 4.9793190189819395e-05,
      "loss": 0.0862,
      "step": 7678,
      "task_loss": 0.22124677896499634
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999305517338018,
      "compression_loss": 0.0,
      "distillation_loss": 0.01819181814789772,
      "epoch": 7.29,
      "learning_rate": 4.979182022801262e-05,
      "loss": 0.0174,
      "step": 7679,
      "task_loss": 0.01074174977838993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999308306421675,
      "compression_loss": 0.0,
      "distillation_loss": 0.045470044016838074,
      "epoch": 7.29,
      "learning_rate": 4.979044576265229e-05,
      "loss": 0.0435,
      "step": 7680,
      "task_loss": 0.025997349992394447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999311088027901,
      "compression_loss": 0.0,
      "distillation_loss": 0.027925997972488403,
      "epoch": 7.29,
      "learning_rate": 4.9789066793988106e-05,
      "loss": 0.0256,
      "step": 7681,
      "task_loss": 0.004869425669312477
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999313862166731,
      "compression_loss": 0.0,
      "distillation_loss": 0.1813095510005951,
      "epoch": 7.3,
      "learning_rate": 4.978768332227054e-05,
      "loss": 0.1863,
      "step": 7682,
      "task_loss": 0.23134742677211761
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999316628848204,
      "compression_loss": 0.0,
      "distillation_loss": 0.015448532067239285,
      "epoch": 7.3,
      "learning_rate": 4.9786295347750936e-05,
      "loss": 0.0143,
      "step": 7683,
      "task_loss": 0.00408821739256382
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999319388082355,
      "compression_loss": 0.0,
      "distillation_loss": 0.0803908184170723,
      "epoch": 7.3,
      "learning_rate": 4.9784902870681406e-05,
      "loss": 0.0909,
      "step": 7684,
      "task_loss": 0.18548454344272614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999322139879224,
      "compression_loss": 0.0,
      "distillation_loss": 0.060036517679691315,
      "epoch": 7.3,
      "learning_rate": 4.97835058913149e-05,
      "loss": 0.0655,
      "step": 7685,
      "task_loss": 0.11482767760753632
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999324884248843,
      "compression_loss": 0.0,
      "distillation_loss": 0.027849406003952026,
      "epoch": 7.3,
      "learning_rate": 4.9782104409905186e-05,
      "loss": 0.0506,
      "step": 7686,
      "task_loss": 0.2550583779811859
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999327621201252,
      "compression_loss": 0.0,
      "distillation_loss": 0.05457767844200134,
      "epoch": 7.3,
      "learning_rate": 4.9780698426706864e-05,
      "loss": 0.0507,
      "step": 7687,
      "task_loss": 0.016233546659350395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999330350746487,
      "compression_loss": 0.0,
      "distillation_loss": 0.09315378218889236,
      "epoch": 7.3,
      "learning_rate": 4.977928794197532e-05,
      "loss": 0.0893,
      "step": 7688,
      "task_loss": 0.05484301596879959
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999333072894586,
      "compression_loss": 0.0,
      "distillation_loss": 0.030269654467701912,
      "epoch": 7.3,
      "learning_rate": 4.9777872955966785e-05,
      "loss": 0.0351,
      "step": 7689,
      "task_loss": 0.0781378373503685
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999335787655584,
      "compression_loss": 0.0,
      "distillation_loss": 0.05370612442493439,
      "epoch": 7.3,
      "learning_rate": 4.97764534689383e-05,
      "loss": 0.0585,
      "step": 7690,
      "task_loss": 0.101178377866745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999338495039519,
      "compression_loss": 0.0,
      "distillation_loss": 0.07610142230987549,
      "epoch": 7.3,
      "learning_rate": 4.977502948114772e-05,
      "loss": 0.0748,
      "step": 7691,
      "task_loss": 0.06338108330965042
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999341195056426,
      "compression_loss": 0.0,
      "distillation_loss": 0.1321769803762436,
      "epoch": 7.3,
      "learning_rate": 4.977360099285371e-05,
      "loss": 0.1249,
      "step": 7692,
      "task_loss": 0.05932076275348663
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999343887716345,
      "compression_loss": 0.0,
      "distillation_loss": 0.02152719907462597,
      "epoch": 7.31,
      "learning_rate": 4.9772168004315765e-05,
      "loss": 0.0272,
      "step": 7693,
      "task_loss": 0.07799746096134186
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799934657302931,
      "compression_loss": 0.0,
      "distillation_loss": 0.021207425743341446,
      "epoch": 7.31,
      "learning_rate": 4.9770730515794204e-05,
      "loss": 0.0265,
      "step": 7694,
      "task_loss": 0.07455786317586899
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999349251005359,
      "compression_loss": 0.0,
      "distillation_loss": 0.0122305266559124,
      "epoch": 7.31,
      "learning_rate": 4.976928852755015e-05,
      "loss": 0.0193,
      "step": 7695,
      "task_loss": 0.08255228400230408
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799935192165453,
      "compression_loss": 0.0,
      "distillation_loss": 0.016809673979878426,
      "epoch": 7.31,
      "learning_rate": 4.976784203984554e-05,
      "loss": 0.016,
      "step": 7696,
      "task_loss": 0.008326346054673195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999354584986857,
      "compression_loss": 0.0,
      "distillation_loss": 0.04126199334859848,
      "epoch": 7.31,
      "learning_rate": 4.976639105294314e-05,
      "loss": 0.0384,
      "step": 7697,
      "task_loss": 0.012333398684859276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999357241012379,
      "compression_loss": 0.0,
      "distillation_loss": 0.1654108762741089,
      "epoch": 7.31,
      "learning_rate": 4.976493556710653e-05,
      "loss": 0.1673,
      "step": 7698,
      "task_loss": 0.18462178111076355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999359889741132,
      "compression_loss": 0.0,
      "distillation_loss": 0.022537576034665108,
      "epoch": 7.31,
      "learning_rate": 4.976347558260011e-05,
      "loss": 0.0206,
      "step": 7699,
      "task_loss": 0.00269523449242115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999362531183153,
      "compression_loss": 0.0,
      "distillation_loss": 0.04379589483141899,
      "epoch": 7.31,
      "learning_rate": 4.976201109968908e-05,
      "loss": 0.0441,
      "step": 7700,
      "task_loss": 0.04727374017238617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999365165348479,
      "compression_loss": 0.0,
      "distillation_loss": 0.06073600426316261,
      "epoch": 7.31,
      "learning_rate": 4.976054211863949e-05,
      "loss": 0.0607,
      "step": 7701,
      "task_loss": 0.05995677784085274
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999367792247146,
      "compression_loss": 0.0,
      "distillation_loss": 0.03459569066762924,
      "epoch": 7.31,
      "learning_rate": 4.9759068639718166e-05,
      "loss": 0.0339,
      "step": 7702,
      "task_loss": 0.027450790628790855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999370411889193,
      "compression_loss": 0.0,
      "distillation_loss": 0.0753156989812851,
      "epoch": 7.32,
      "learning_rate": 4.975759066319278e-05,
      "loss": 0.0687,
      "step": 7703,
      "task_loss": 0.00872378796339035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999373024284654,
      "compression_loss": 0.0,
      "distillation_loss": 0.029609953984618187,
      "epoch": 7.32,
      "learning_rate": 4.9756108189331825e-05,
      "loss": 0.0271,
      "step": 7704,
      "task_loss": 0.004311494529247284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999375629443568,
      "compression_loss": 0.0,
      "distillation_loss": 0.0182811226695776,
      "epoch": 7.32,
      "learning_rate": 4.975462121840458e-05,
      "loss": 0.0215,
      "step": 7705,
      "task_loss": 0.05089661478996277
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999378227375971,
      "compression_loss": 0.0,
      "distillation_loss": 0.027917640283703804,
      "epoch": 7.32,
      "learning_rate": 4.975312975068118e-05,
      "loss": 0.0381,
      "step": 7706,
      "task_loss": 0.1298721432685852
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999380818091898,
      "compression_loss": 0.0,
      "distillation_loss": 0.06191258504986763,
      "epoch": 7.32,
      "learning_rate": 4.975163378643255e-05,
      "loss": 0.058,
      "step": 7707,
      "task_loss": 0.02266554906964302
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999383401601389,
      "compression_loss": 0.0,
      "distillation_loss": 0.02043573185801506,
      "epoch": 7.32,
      "learning_rate": 4.975013332593044e-05,
      "loss": 0.021,
      "step": 7708,
      "task_loss": 0.026547657325863838
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999385977914479,
      "compression_loss": 0.0,
      "distillation_loss": 0.018625617027282715,
      "epoch": 7.32,
      "learning_rate": 4.97486283694474e-05,
      "loss": 0.0172,
      "step": 7709,
      "task_loss": 0.004806183278560638
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999388547041205,
      "compression_loss": 0.0,
      "distillation_loss": 0.11342775821685791,
      "epoch": 7.32,
      "learning_rate": 4.974711891725684e-05,
      "loss": 0.1113,
      "step": 7710,
      "task_loss": 0.09240047633647919
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999391108991605,
      "compression_loss": 0.0,
      "distillation_loss": 0.025932233780622482,
      "epoch": 7.32,
      "learning_rate": 4.9745604969632934e-05,
      "loss": 0.0389,
      "step": 7711,
      "task_loss": 0.15590538084506989
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999393663775713,
      "compression_loss": 0.0,
      "distillation_loss": 0.15824320912361145,
      "epoch": 7.32,
      "learning_rate": 4.974408652685072e-05,
      "loss": 0.1603,
      "step": 7712,
      "task_loss": 0.1786690652370453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799939621140357,
      "compression_loss": 0.0,
      "distillation_loss": 0.1327803134918213,
      "epoch": 7.32,
      "learning_rate": 4.974256358918601e-05,
      "loss": 0.1301,
      "step": 7713,
      "task_loss": 0.1060781478881836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999398751885209,
      "compression_loss": 0.0,
      "distillation_loss": 0.024752333760261536,
      "epoch": 7.33,
      "learning_rate": 4.9741036156915464e-05,
      "loss": 0.0296,
      "step": 7714,
      "task_loss": 0.07322291284799576
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799940128523067,
      "compression_loss": 0.0,
      "distillation_loss": 0.04586133733391762,
      "epoch": 7.33,
      "learning_rate": 4.973950423031655e-05,
      "loss": 0.0429,
      "step": 7715,
      "task_loss": 0.016736658290028572
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999403811449985,
      "compression_loss": 0.0,
      "distillation_loss": 0.014082120731472969,
      "epoch": 7.33,
      "learning_rate": 4.9737967809667546e-05,
      "loss": 0.0198,
      "step": 7716,
      "task_loss": 0.07149508595466614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999406330553197,
      "compression_loss": 0.0,
      "distillation_loss": 0.029716944321990013,
      "epoch": 7.33,
      "learning_rate": 4.9736426895247545e-05,
      "loss": 0.0355,
      "step": 7717,
      "task_loss": 0.08792907744646072
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999408842550338,
      "compression_loss": 0.0,
      "distillation_loss": 0.012465763837099075,
      "epoch": 7.33,
      "learning_rate": 4.973488148733647e-05,
      "loss": 0.0118,
      "step": 7718,
      "task_loss": 0.005948711186647415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999411347451447,
      "compression_loss": 0.0,
      "distillation_loss": 0.028147976845502853,
      "epoch": 7.33,
      "learning_rate": 4.973333158621505e-05,
      "loss": 0.0256,
      "step": 7719,
      "task_loss": 0.0030502378940582275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999413845266561,
      "compression_loss": 0.0,
      "distillation_loss": 0.26714271306991577,
      "epoch": 7.33,
      "learning_rate": 4.973177719216483e-05,
      "loss": 0.2669,
      "step": 7720,
      "task_loss": 0.26459577679634094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999416336005716,
      "compression_loss": 0.0,
      "distillation_loss": 0.0716482624411583,
      "epoch": 7.33,
      "learning_rate": 4.973021830546817e-05,
      "loss": 0.0705,
      "step": 7721,
      "task_loss": 0.06003544479608536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999418819678948,
      "compression_loss": 0.0,
      "distillation_loss": 0.03609579801559448,
      "epoch": 7.33,
      "learning_rate": 4.972865492640826e-05,
      "loss": 0.0456,
      "step": 7722,
      "task_loss": 0.13090340793132782
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999421296296296,
      "compression_loss": 0.0,
      "distillation_loss": 0.09618473052978516,
      "epoch": 7.33,
      "learning_rate": 4.972708705526908e-05,
      "loss": 0.1099,
      "step": 7723,
      "task_loss": 0.23324905335903168
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999423765867796,
      "compression_loss": 0.0,
      "distillation_loss": 0.04280245676636696,
      "epoch": 7.34,
      "learning_rate": 4.972551469233545e-05,
      "loss": 0.0415,
      "step": 7724,
      "task_loss": 0.029460199177265167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999426228403484,
      "compression_loss": 0.0,
      "distillation_loss": 0.2212740182876587,
      "epoch": 7.34,
      "learning_rate": 4.9723937837892996e-05,
      "loss": 0.2304,
      "step": 7725,
      "task_loss": 0.3125489354133606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999428683913398,
      "compression_loss": 0.0,
      "distillation_loss": 0.04743258282542229,
      "epoch": 7.34,
      "learning_rate": 4.972235649222817e-05,
      "loss": 0.0538,
      "step": 7726,
      "task_loss": 0.11064188182353973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999431132407573,
      "compression_loss": 0.0,
      "distillation_loss": 0.04925276339054108,
      "epoch": 7.34,
      "learning_rate": 4.972077065562821e-05,
      "loss": 0.0582,
      "step": 7727,
      "task_loss": 0.13823464512825012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999433573896048,
      "compression_loss": 0.0,
      "distillation_loss": 0.05824163183569908,
      "epoch": 7.34,
      "learning_rate": 4.971918032838122e-05,
      "loss": 0.0562,
      "step": 7728,
      "task_loss": 0.03779482841491699
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999436008388858,
      "compression_loss": 0.0,
      "distillation_loss": 0.08110888302326202,
      "epoch": 7.34,
      "learning_rate": 4.9717585510776065e-05,
      "loss": 0.08,
      "step": 7729,
      "task_loss": 0.06993526965379715
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999438435896041,
      "compression_loss": 0.0,
      "distillation_loss": 0.026565806940197945,
      "epoch": 7.34,
      "learning_rate": 4.971598620310246e-05,
      "loss": 0.0247,
      "step": 7730,
      "task_loss": 0.007519997656345367
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999440856427634,
      "compression_loss": 0.0,
      "distillation_loss": 0.03230992704629898,
      "epoch": 7.34,
      "learning_rate": 4.9714382405650926e-05,
      "loss": 0.0364,
      "step": 7731,
      "task_loss": 0.07325948029756546
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999443269993672,
      "compression_loss": 0.0,
      "distillation_loss": 0.020386017858982086,
      "epoch": 7.34,
      "learning_rate": 4.971277411871281e-05,
      "loss": 0.0193,
      "step": 7732,
      "task_loss": 0.009038899093866348
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999445676604194,
      "compression_loss": 0.0,
      "distillation_loss": 0.07571500539779663,
      "epoch": 7.34,
      "learning_rate": 4.971116134258025e-05,
      "loss": 0.0778,
      "step": 7733,
      "task_loss": 0.09688396751880646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999448076269235,
      "compression_loss": 0.0,
      "distillation_loss": 0.051084429025650024,
      "epoch": 7.34,
      "learning_rate": 4.9709544077546235e-05,
      "loss": 0.0616,
      "step": 7734,
      "task_loss": 0.1558133214712143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999450468998834,
      "compression_loss": 0.0,
      "distillation_loss": 0.05314343795180321,
      "epoch": 7.35,
      "learning_rate": 4.9707922323904524e-05,
      "loss": 0.0731,
      "step": 7735,
      "task_loss": 0.25305381417274475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999452854803026,
      "compression_loss": 0.0,
      "distillation_loss": 0.02783917263150215,
      "epoch": 7.35,
      "learning_rate": 4.9706296081949724e-05,
      "loss": 0.0271,
      "step": 7736,
      "task_loss": 0.020391501486301422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999455233691849,
      "compression_loss": 0.0,
      "distillation_loss": 0.13770464062690735,
      "epoch": 7.35,
      "learning_rate": 4.9704665351977266e-05,
      "loss": 0.1352,
      "step": 7737,
      "task_loss": 0.11255469173192978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999457605675339,
      "compression_loss": 0.0,
      "distillation_loss": 0.06904464215040207,
      "epoch": 7.35,
      "learning_rate": 4.9703030134283356e-05,
      "loss": 0.063,
      "step": 7738,
      "task_loss": 0.008208906278014183
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999459970763533,
      "compression_loss": 0.0,
      "distillation_loss": 0.14025121927261353,
      "epoch": 7.35,
      "learning_rate": 4.970139042916506e-05,
      "loss": 0.1517,
      "step": 7739,
      "task_loss": 0.25468742847442627
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999462328966467,
      "compression_loss": 0.0,
      "distillation_loss": 0.014299793168902397,
      "epoch": 7.35,
      "learning_rate": 4.969974623692023e-05,
      "loss": 0.0224,
      "step": 7740,
      "task_loss": 0.09552035480737686
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999464680294179,
      "compression_loss": 0.0,
      "distillation_loss": 0.050735875964164734,
      "epoch": 7.35,
      "learning_rate": 4.969809755784753e-05,
      "loss": 0.046,
      "step": 7741,
      "task_loss": 0.003554748371243477
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999467024756706,
      "compression_loss": 0.0,
      "distillation_loss": 0.11840305477380753,
      "epoch": 7.35,
      "learning_rate": 4.969644439224647e-05,
      "loss": 0.1174,
      "step": 7742,
      "task_loss": 0.10836636275053024
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999469362364083,
      "compression_loss": 0.0,
      "distillation_loss": 0.03614204376935959,
      "epoch": 7.35,
      "learning_rate": 4.969478674041735e-05,
      "loss": 0.0386,
      "step": 7743,
      "task_loss": 0.061211228370666504
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799947169312635,
      "compression_loss": 0.0,
      "distillation_loss": 0.054497864097356796,
      "epoch": 7.35,
      "learning_rate": 4.969312460266128e-05,
      "loss": 0.0529,
      "step": 7744,
      "task_loss": 0.0389748215675354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799947401705354,
      "compression_loss": 0.0,
      "distillation_loss": 0.0671176165342331,
      "epoch": 7.36,
      "learning_rate": 4.969145797928021e-05,
      "loss": 0.0707,
      "step": 7745,
      "task_loss": 0.10274288058280945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999476334155693,
      "compression_loss": 0.0,
      "distillation_loss": 0.03849276900291443,
      "epoch": 7.36,
      "learning_rate": 4.968978687057687e-05,
      "loss": 0.0368,
      "step": 7746,
      "task_loss": 0.02152402326464653
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999478644442845,
      "compression_loss": 0.0,
      "distillation_loss": 0.0317964106798172,
      "epoch": 7.36,
      "learning_rate": 4.9688111276854846e-05,
      "loss": 0.036,
      "step": 7747,
      "task_loss": 0.0738469660282135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999480947925032,
      "compression_loss": 0.0,
      "distillation_loss": 0.10060693323612213,
      "epoch": 7.36,
      "learning_rate": 4.9686431198418515e-05,
      "loss": 0.0981,
      "step": 7748,
      "task_loss": 0.07562766224145889
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799948324461229,
      "compression_loss": 0.0,
      "distillation_loss": 0.17574167251586914,
      "epoch": 7.36,
      "learning_rate": 4.968474663557306e-05,
      "loss": 0.1741,
      "step": 7749,
      "task_loss": 0.1597302258014679
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999485534514659,
      "compression_loss": 0.0,
      "distillation_loss": 0.05716216191649437,
      "epoch": 7.36,
      "learning_rate": 4.9683057588624494e-05,
      "loss": 0.0604,
      "step": 7750,
      "task_loss": 0.08970184624195099
    },
    {
      "epoch": 7.36,
      "eval_accuracy": 0.8761467889908257,
      "eval_loss": 0.5915129780769348,
      "eval_runtime": 18.0311,
      "eval_samples_per_second": 48.361,
      "eval_steps_per_second": 6.045,
      "step": 7750
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999487817642172,
      "compression_loss": 0.0,
      "distillation_loss": 0.0786730945110321,
      "epoch": 7.36,
      "learning_rate": 4.968136405787964e-05,
      "loss": 0.0764,
      "step": 7751,
      "task_loss": 0.05567537993192673
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999490094004869,
      "compression_loss": 0.0,
      "distillation_loss": 0.12053617835044861,
      "epoch": 7.36,
      "learning_rate": 4.967966604364614e-05,
      "loss": 0.12,
      "step": 7752,
      "task_loss": 0.11506214737892151
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999492363612786,
      "compression_loss": 0.0,
      "distillation_loss": 0.13453137874603271,
      "epoch": 7.36,
      "learning_rate": 4.9677963546232445e-05,
      "loss": 0.1363,
      "step": 7753,
      "task_loss": 0.152423694729805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999494626475958,
      "compression_loss": 0.0,
      "distillation_loss": 0.05290111154317856,
      "epoch": 7.36,
      "learning_rate": 4.967625656594782e-05,
      "loss": 0.0494,
      "step": 7754,
      "task_loss": 0.018385985866189003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999496882604423,
      "compression_loss": 0.0,
      "distillation_loss": 0.11135755479335785,
      "epoch": 7.36,
      "learning_rate": 4.967454510310235e-05,
      "loss": 0.1068,
      "step": 7755,
      "task_loss": 0.06577824056148529
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999499132008219,
      "compression_loss": 0.0,
      "distillation_loss": 0.03572801128029823,
      "epoch": 7.37,
      "learning_rate": 4.967282915800693e-05,
      "loss": 0.0332,
      "step": 7756,
      "task_loss": 0.01009390503168106
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999501374697381,
      "compression_loss": 0.0,
      "distillation_loss": 0.13653673231601715,
      "epoch": 7.37,
      "learning_rate": 4.9671108730973274e-05,
      "loss": 0.1311,
      "step": 7757,
      "task_loss": 0.08257921040058136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999503610681947,
      "compression_loss": 0.0,
      "distillation_loss": 0.01905696839094162,
      "epoch": 7.37,
      "learning_rate": 4.9669383822313886e-05,
      "loss": 0.0178,
      "step": 7758,
      "task_loss": 0.006181072443723679
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999505839971953,
      "compression_loss": 0.0,
      "distillation_loss": 0.051291391253471375,
      "epoch": 7.37,
      "learning_rate": 4.966765443234212e-05,
      "loss": 0.0508,
      "step": 7759,
      "task_loss": 0.04657955840229988
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999508062577437,
      "compression_loss": 0.0,
      "distillation_loss": 0.025289995595812798,
      "epoch": 7.37,
      "learning_rate": 4.966592056137213e-05,
      "loss": 0.0322,
      "step": 7760,
      "task_loss": 0.09458184987306595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999510278508435,
      "compression_loss": 0.0,
      "distillation_loss": 0.022838320583105087,
      "epoch": 7.37,
      "learning_rate": 4.966418220971888e-05,
      "loss": 0.031,
      "step": 7761,
      "task_loss": 0.10419056564569473
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999512487774983,
      "compression_loss": 0.0,
      "distillation_loss": 0.015206392854452133,
      "epoch": 7.37,
      "learning_rate": 4.9662439377698145e-05,
      "loss": 0.0172,
      "step": 7762,
      "task_loss": 0.03513802960515022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999514690387121,
      "compression_loss": 0.0,
      "distillation_loss": 0.07149576395750046,
      "epoch": 7.37,
      "learning_rate": 4.9660692065626515e-05,
      "loss": 0.0657,
      "step": 7763,
      "task_loss": 0.013111604377627373
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999516886354882,
      "compression_loss": 0.0,
      "distillation_loss": 0.0190842617303133,
      "epoch": 7.37,
      "learning_rate": 4.965894027382141e-05,
      "loss": 0.0176,
      "step": 7764,
      "task_loss": 0.00447353720664978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999519075688304,
      "compression_loss": 0.0,
      "distillation_loss": 0.027284346520900726,
      "epoch": 7.37,
      "learning_rate": 4.965718400260105e-05,
      "loss": 0.0253,
      "step": 7765,
      "task_loss": 0.007609104737639427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999521258397424,
      "compression_loss": 0.0,
      "distillation_loss": 0.05418026074767113,
      "epoch": 7.38,
      "learning_rate": 4.965542325228446e-05,
      "loss": 0.0654,
      "step": 7766,
      "task_loss": 0.16616730391979218
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999523434492279,
      "compression_loss": 0.0,
      "distillation_loss": 0.09797915816307068,
      "epoch": 7.38,
      "learning_rate": 4.96536580231915e-05,
      "loss": 0.106,
      "step": 7767,
      "task_loss": 0.17831069231033325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999525603982907,
      "compression_loss": 0.0,
      "distillation_loss": 0.09306447952985764,
      "epoch": 7.38,
      "learning_rate": 4.9651888315642815e-05,
      "loss": 0.0886,
      "step": 7768,
      "task_loss": 0.04870466887950897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999527766879343,
      "compression_loss": 0.0,
      "distillation_loss": 0.04752252995967865,
      "epoch": 7.38,
      "learning_rate": 4.96501141299599e-05,
      "loss": 0.0523,
      "step": 7769,
      "task_loss": 0.09561814367771149
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999529923191625,
      "compression_loss": 0.0,
      "distillation_loss": 0.05140899494290352,
      "epoch": 7.38,
      "learning_rate": 4.9648335466465035e-05,
      "loss": 0.0529,
      "step": 7770,
      "task_loss": 0.06599454581737518
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999532072929788,
      "compression_loss": 0.0,
      "distillation_loss": 0.049043748527765274,
      "epoch": 7.38,
      "learning_rate": 4.964655232548133e-05,
      "loss": 0.0707,
      "step": 7771,
      "task_loss": 0.2652056813240051
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999534216103871,
      "compression_loss": 0.0,
      "distillation_loss": 0.08960321545600891,
      "epoch": 7.38,
      "learning_rate": 4.964476470733269e-05,
      "loss": 0.093,
      "step": 7772,
      "task_loss": 0.12396573275327682
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799953635272391,
      "compression_loss": 0.0,
      "distillation_loss": 0.028217440471053123,
      "epoch": 7.38,
      "learning_rate": 4.964297261234385e-05,
      "loss": 0.0295,
      "step": 7773,
      "task_loss": 0.0411832258105278
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999538482799942,
      "compression_loss": 0.0,
      "distillation_loss": 0.03329095244407654,
      "epoch": 7.38,
      "learning_rate": 4.964117604084036e-05,
      "loss": 0.0398,
      "step": 7774,
      "task_loss": 0.09877783805131912
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999540606342003,
      "compression_loss": 0.0,
      "distillation_loss": 0.09195288270711899,
      "epoch": 7.38,
      "learning_rate": 4.963937499314857e-05,
      "loss": 0.1022,
      "step": 7775,
      "task_loss": 0.1942749172449112
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799954272336013,
      "compression_loss": 0.0,
      "distillation_loss": 0.036124154925346375,
      "epoch": 7.38,
      "learning_rate": 4.963756946959564e-05,
      "loss": 0.0386,
      "step": 7776,
      "task_loss": 0.060537442564964294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799954483386436,
      "compression_loss": 0.0,
      "distillation_loss": 0.04166106879711151,
      "epoch": 7.39,
      "learning_rate": 4.9635759470509554e-05,
      "loss": 0.0392,
      "step": 7777,
      "task_loss": 0.017511041834950447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999546937864732,
      "compression_loss": 0.0,
      "distillation_loss": 0.03311231732368469,
      "epoch": 7.39,
      "learning_rate": 4.9633944996219125e-05,
      "loss": 0.0324,
      "step": 7778,
      "task_loss": 0.025931518524885178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799954903537128,
      "compression_loss": 0.0,
      "distillation_loss": 0.11695058643817902,
      "epoch": 7.39,
      "learning_rate": 4.9632126047053954e-05,
      "loss": 0.1216,
      "step": 7779,
      "task_loss": 0.1629902720451355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999551126394041,
      "compression_loss": 0.0,
      "distillation_loss": 0.12245595455169678,
      "epoch": 7.39,
      "learning_rate": 4.963030262334445e-05,
      "loss": 0.1158,
      "step": 7780,
      "task_loss": 0.05568038299679756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999553210943052,
      "compression_loss": 0.0,
      "distillation_loss": 0.027472712099552155,
      "epoch": 7.39,
      "learning_rate": 4.962847472542185e-05,
      "loss": 0.031,
      "step": 7781,
      "task_loss": 0.062434788793325424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999555289028352,
      "compression_loss": 0.0,
      "distillation_loss": 0.04194699972867966,
      "epoch": 7.39,
      "learning_rate": 4.96266423536182e-05,
      "loss": 0.0484,
      "step": 7782,
      "task_loss": 0.10626760870218277
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999557360659975,
      "compression_loss": 0.0,
      "distillation_loss": 0.05600655823945999,
      "epoch": 7.39,
      "learning_rate": 4.9624805508266375e-05,
      "loss": 0.0603,
      "step": 7783,
      "task_loss": 0.09906087070703506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799955942584796,
      "compression_loss": 0.0,
      "distillation_loss": 0.03841715306043625,
      "epoch": 7.39,
      "learning_rate": 4.9622964189700026e-05,
      "loss": 0.0373,
      "step": 7784,
      "task_loss": 0.027386927977204323
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999561484602341,
      "compression_loss": 0.0,
      "distillation_loss": 0.033997707068920135,
      "epoch": 7.39,
      "learning_rate": 4.962111839825365e-05,
      "loss": 0.0323,
      "step": 7785,
      "task_loss": 0.016779828816652298
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999563536933157,
      "compression_loss": 0.0,
      "distillation_loss": 0.05404585599899292,
      "epoch": 7.39,
      "learning_rate": 4.961926813426254e-05,
      "loss": 0.0535,
      "step": 7786,
      "task_loss": 0.04876035451889038
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999565582850446,
      "compression_loss": 0.0,
      "distillation_loss": 0.03144819289445877,
      "epoch": 7.4,
      "learning_rate": 4.9617413398062814e-05,
      "loss": 0.0358,
      "step": 7787,
      "task_loss": 0.07468868046998978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999567622364242,
      "compression_loss": 0.0,
      "distillation_loss": 0.09421627223491669,
      "epoch": 7.4,
      "learning_rate": 4.9615554189991374e-05,
      "loss": 0.0962,
      "step": 7788,
      "task_loss": 0.11395429074764252
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999569655484584,
      "compression_loss": 0.0,
      "distillation_loss": 0.05254974216222763,
      "epoch": 7.4,
      "learning_rate": 4.9613690510385965e-05,
      "loss": 0.0524,
      "step": 7789,
      "task_loss": 0.051460277289152145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999571682221507,
      "compression_loss": 0.0,
      "distillation_loss": 0.1056189015507698,
      "epoch": 7.4,
      "learning_rate": 4.961182235958515e-05,
      "loss": 0.1065,
      "step": 7790,
      "task_loss": 0.11400679498910904
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999573702585049,
      "compression_loss": 0.0,
      "distillation_loss": 0.04173080623149872,
      "epoch": 7.4,
      "learning_rate": 4.9609949737928254e-05,
      "loss": 0.0572,
      "step": 7791,
      "task_loss": 0.19677627086639404
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999575716585247,
      "compression_loss": 0.0,
      "distillation_loss": 0.021544422954320908,
      "epoch": 7.4,
      "learning_rate": 4.9608072645755476e-05,
      "loss": 0.0198,
      "step": 7792,
      "task_loss": 0.004072500392794609
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999577724232136,
      "compression_loss": 0.0,
      "distillation_loss": 0.06151081249117851,
      "epoch": 7.4,
      "learning_rate": 4.960619108340778e-05,
      "loss": 0.058,
      "step": 7793,
      "task_loss": 0.02649463526904583
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999579725535755,
      "compression_loss": 0.0,
      "distillation_loss": 0.07832042872905731,
      "epoch": 7.4,
      "learning_rate": 4.9604305051226976e-05,
      "loss": 0.0742,
      "step": 7794,
      "task_loss": 0.03758959844708443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799958172050614,
      "compression_loss": 0.0,
      "distillation_loss": 0.058107852935791016,
      "epoch": 7.4,
      "learning_rate": 4.960241454955566e-05,
      "loss": 0.0527,
      "step": 7795,
      "task_loss": 0.004193047061562538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999583709153328,
      "compression_loss": 0.0,
      "distillation_loss": 0.025543441995978355,
      "epoch": 7.4,
      "learning_rate": 4.960051957873725e-05,
      "loss": 0.0301,
      "step": 7796,
      "task_loss": 0.07116261124610901
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999585691487355,
      "compression_loss": 0.0,
      "distillation_loss": 0.09677424281835556,
      "epoch": 7.4,
      "learning_rate": 4.959862013911599e-05,
      "loss": 0.0951,
      "step": 7797,
      "task_loss": 0.07997578382492065
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799958766751826,
      "compression_loss": 0.0,
      "distillation_loss": 0.03689539432525635,
      "epoch": 7.41,
      "learning_rate": 4.959671623103691e-05,
      "loss": 0.0423,
      "step": 7798,
      "task_loss": 0.09053385257720947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999589637256077,
      "compression_loss": 0.0,
      "distillation_loss": 0.09475536644458771,
      "epoch": 7.41,
      "learning_rate": 4.959480785484587e-05,
      "loss": 0.1033,
      "step": 7799,
      "task_loss": 0.17996615171432495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999591600710844,
      "compression_loss": 0.0,
      "distillation_loss": 0.10527956485748291,
      "epoch": 7.41,
      "learning_rate": 4.959289501088953e-05,
      "loss": 0.0973,
      "step": 7800,
      "task_loss": 0.025774430483579636
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999593557892598,
      "compression_loss": 0.0,
      "distillation_loss": 0.04095172882080078,
      "epoch": 7.41,
      "learning_rate": 4.9590977699515374e-05,
      "loss": 0.0552,
      "step": 7801,
      "task_loss": 0.18330860137939453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999595508811376,
      "compression_loss": 0.0,
      "distillation_loss": 0.03950580954551697,
      "epoch": 7.41,
      "learning_rate": 4.958905592107168e-05,
      "loss": 0.0592,
      "step": 7802,
      "task_loss": 0.2363201081752777
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999597453477214,
      "compression_loss": 0.0,
      "distillation_loss": 0.1426384150981903,
      "epoch": 7.41,
      "learning_rate": 4.958712967590756e-05,
      "loss": 0.1449,
      "step": 7803,
      "task_loss": 0.16487504541873932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999599391900151,
      "compression_loss": 0.0,
      "distillation_loss": 0.03035861626267433,
      "epoch": 7.41,
      "learning_rate": 4.9585198964372925e-05,
      "loss": 0.0333,
      "step": 7804,
      "task_loss": 0.060261476784944534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999601324090221,
      "compression_loss": 0.0,
      "distillation_loss": 0.0202895849943161,
      "epoch": 7.41,
      "learning_rate": 4.958326378681849e-05,
      "loss": 0.0284,
      "step": 7805,
      "task_loss": 0.10182420909404755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999603250057462,
      "compression_loss": 0.0,
      "distillation_loss": 0.04021252319216728,
      "epoch": 7.41,
      "learning_rate": 4.958132414359579e-05,
      "loss": 0.0507,
      "step": 7806,
      "task_loss": 0.14528432488441467
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999605169811911,
      "compression_loss": 0.0,
      "distillation_loss": 0.14047583937644958,
      "epoch": 7.41,
      "learning_rate": 4.957938003505718e-05,
      "loss": 0.1381,
      "step": 7807,
      "task_loss": 0.1167716383934021
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999607083363605,
      "compression_loss": 0.0,
      "distillation_loss": 0.06726957857608795,
      "epoch": 7.42,
      "learning_rate": 4.957743146155581e-05,
      "loss": 0.0725,
      "step": 7808,
      "task_loss": 0.11913690716028214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799960899072258,
      "compression_loss": 0.0,
      "distillation_loss": 0.02605987712740898,
      "epoch": 7.42,
      "learning_rate": 4.9575478423445655e-05,
      "loss": 0.024,
      "step": 7809,
      "task_loss": 0.005071789026260376
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999610891898873,
      "compression_loss": 0.0,
      "distillation_loss": 0.04089764878153801,
      "epoch": 7.42,
      "learning_rate": 4.957352092108148e-05,
      "loss": 0.0429,
      "step": 7810,
      "task_loss": 0.06113029643893242
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999612786902521,
      "compression_loss": 0.0,
      "distillation_loss": 0.014122812077403069,
      "epoch": 7.42,
      "learning_rate": 4.957155895481889e-05,
      "loss": 0.0132,
      "step": 7811,
      "task_loss": 0.005167461931705475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999614675743562,
      "compression_loss": 0.0,
      "distillation_loss": 0.013877642340958118,
      "epoch": 7.42,
      "learning_rate": 4.956959252501426e-05,
      "loss": 0.0172,
      "step": 7812,
      "task_loss": 0.047237932682037354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999616558432031,
      "compression_loss": 0.0,
      "distillation_loss": 0.04274125397205353,
      "epoch": 7.42,
      "learning_rate": 4.956762163202484e-05,
      "loss": 0.0458,
      "step": 7813,
      "task_loss": 0.07380035519599915
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999618434977966,
      "compression_loss": 0.0,
      "distillation_loss": 0.18687260150909424,
      "epoch": 7.42,
      "learning_rate": 4.956564627620862e-05,
      "loss": 0.1937,
      "step": 7814,
      "task_loss": 0.25557661056518555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999620305391403,
      "compression_loss": 0.0,
      "distillation_loss": 0.136896014213562,
      "epoch": 7.42,
      "learning_rate": 4.956366645792445e-05,
      "loss": 0.1269,
      "step": 7815,
      "task_loss": 0.03715690225362778
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799962216968238,
      "compression_loss": 0.0,
      "distillation_loss": 0.011462707072496414,
      "epoch": 7.42,
      "learning_rate": 4.956168217753197e-05,
      "loss": 0.0107,
      "step": 7816,
      "task_loss": 0.004216820001602173
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999624027860932,
      "compression_loss": 0.0,
      "distillation_loss": 0.030813174322247505,
      "epoch": 7.42,
      "learning_rate": 4.955969343539162e-05,
      "loss": 0.0384,
      "step": 7817,
      "task_loss": 0.10682034492492676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999625879937098,
      "compression_loss": 0.0,
      "distillation_loss": 0.0257425494492054,
      "epoch": 7.42,
      "learning_rate": 4.955770023186469e-05,
      "loss": 0.024,
      "step": 7818,
      "task_loss": 0.007959723472595215
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999627725920913,
      "compression_loss": 0.0,
      "distillation_loss": 0.10705772042274475,
      "epoch": 7.43,
      "learning_rate": 4.9555702567313235e-05,
      "loss": 0.1132,
      "step": 7819,
      "task_loss": 0.1683865487575531
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999629565822415,
      "compression_loss": 0.0,
      "distillation_loss": 0.07967795431613922,
      "epoch": 7.43,
      "learning_rate": 4.9553700442100146e-05,
      "loss": 0.0802,
      "step": 7820,
      "task_loss": 0.08508510887622833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799963139965164,
      "compression_loss": 0.0,
      "distillation_loss": 0.14191925525665283,
      "epoch": 7.43,
      "learning_rate": 4.955169385658912e-05,
      "loss": 0.1483,
      "step": 7821,
      "task_loss": 0.20544429123401642
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999633227418624,
      "compression_loss": 0.0,
      "distillation_loss": 0.025436338037252426,
      "epoch": 7.43,
      "learning_rate": 4.954968281114467e-05,
      "loss": 0.0262,
      "step": 7822,
      "task_loss": 0.03303007036447525
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999635049133407,
      "compression_loss": 0.0,
      "distillation_loss": 0.025042004883289337,
      "epoch": 7.43,
      "learning_rate": 4.9547667306132096e-05,
      "loss": 0.0274,
      "step": 7823,
      "task_loss": 0.048622217029333115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999636864806022,
      "compression_loss": 0.0,
      "distillation_loss": 0.04454713687300682,
      "epoch": 7.43,
      "learning_rate": 4.954564734191753e-05,
      "loss": 0.0541,
      "step": 7824,
      "task_loss": 0.13989417254924774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799963867444651,
      "compression_loss": 0.0,
      "distillation_loss": 0.037809595465660095,
      "epoch": 7.43,
      "learning_rate": 4.9543622918867926e-05,
      "loss": 0.0451,
      "step": 7825,
      "task_loss": 0.11072047799825668
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999640478064903,
      "compression_loss": 0.0,
      "distillation_loss": 0.10665163397789001,
      "epoch": 7.43,
      "learning_rate": 4.9541594037351e-05,
      "loss": 0.1153,
      "step": 7826,
      "task_loss": 0.19337314367294312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999642275671242,
      "compression_loss": 0.0,
      "distillation_loss": 0.07981669902801514,
      "epoch": 7.43,
      "learning_rate": 4.953956069773534e-05,
      "loss": 0.0826,
      "step": 7827,
      "task_loss": 0.10715599358081818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999644067275561,
      "compression_loss": 0.0,
      "distillation_loss": 0.026072172448039055,
      "epoch": 7.43,
      "learning_rate": 4.953752290039028e-05,
      "loss": 0.0309,
      "step": 7828,
      "task_loss": 0.07401913404464722
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999645852887898,
      "compression_loss": 0.0,
      "distillation_loss": 0.05162549391388893,
      "epoch": 7.43,
      "learning_rate": 4.953548064568602e-05,
      "loss": 0.0483,
      "step": 7829,
      "task_loss": 0.018601059913635254
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799964763251829,
      "compression_loss": 0.0,
      "distillation_loss": 0.03277206048369408,
      "epoch": 7.44,
      "learning_rate": 4.953343393399354e-05,
      "loss": 0.0379,
      "step": 7830,
      "task_loss": 0.08411306887865067
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999649406176773,
      "compression_loss": 0.0,
      "distillation_loss": 0.023461688309907913,
      "epoch": 7.44,
      "learning_rate": 4.953138276568462e-05,
      "loss": 0.0346,
      "step": 7831,
      "task_loss": 0.13488860428333282
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999651173873386,
      "compression_loss": 0.0,
      "distillation_loss": 0.20481958985328674,
      "epoch": 7.44,
      "learning_rate": 4.952932714113188e-05,
      "loss": 0.2078,
      "step": 7832,
      "task_loss": 0.2342824935913086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999652935618162,
      "compression_loss": 0.0,
      "distillation_loss": 0.05175776779651642,
      "epoch": 7.44,
      "learning_rate": 4.9527267060708734e-05,
      "loss": 0.0608,
      "step": 7833,
      "task_loss": 0.14181137084960938
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999654691421142,
      "compression_loss": 0.0,
      "distillation_loss": 0.1361636519432068,
      "epoch": 7.44,
      "learning_rate": 4.9525202524789397e-05,
      "loss": 0.1285,
      "step": 7834,
      "task_loss": 0.059967104345560074
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799965644129236,
      "compression_loss": 0.0,
      "distillation_loss": 0.03751971200108528,
      "epoch": 7.44,
      "learning_rate": 4.952313353374891e-05,
      "loss": 0.0376,
      "step": 7835,
      "task_loss": 0.03848748654127121
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999658185241852,
      "compression_loss": 0.0,
      "distillation_loss": 0.03644530102610588,
      "epoch": 7.44,
      "learning_rate": 4.952106008796311e-05,
      "loss": 0.0375,
      "step": 7836,
      "task_loss": 0.04672703891992569
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799965992327966,
      "compression_loss": 0.0,
      "distillation_loss": 0.03545859456062317,
      "epoch": 7.44,
      "learning_rate": 4.9518982187808653e-05,
      "loss": 0.036,
      "step": 7837,
      "task_loss": 0.040626343339681625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999661655415814,
      "compression_loss": 0.0,
      "distillation_loss": 0.04039856046438217,
      "epoch": 7.44,
      "learning_rate": 4.9516899833663e-05,
      "loss": 0.0369,
      "step": 7838,
      "task_loss": 0.005691641941666603
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999663381660356,
      "compression_loss": 0.0,
      "distillation_loss": 0.019872594624757767,
      "epoch": 7.44,
      "learning_rate": 4.9514813025904413e-05,
      "loss": 0.0265,
      "step": 7839,
      "task_loss": 0.0859319418668747
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799966510202332,
      "compression_loss": 0.0,
      "distillation_loss": 0.024417951703071594,
      "epoch": 7.45,
      "learning_rate": 4.951272176491197e-05,
      "loss": 0.0301,
      "step": 7840,
      "task_loss": 0.08107300847768784
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999666816514744,
      "compression_loss": 0.0,
      "distillation_loss": 0.03704302757978439,
      "epoch": 7.45,
      "learning_rate": 4.951062605106557e-05,
      "loss": 0.0485,
      "step": 7841,
      "task_loss": 0.15177100896835327
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999668525144665,
      "compression_loss": 0.0,
      "distillation_loss": 0.04666809365153313,
      "epoch": 7.45,
      "learning_rate": 4.950852588474591e-05,
      "loss": 0.0444,
      "step": 7842,
      "task_loss": 0.024120669811964035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999670227923119,
      "compression_loss": 0.0,
      "distillation_loss": 0.02970319800078869,
      "epoch": 7.45,
      "learning_rate": 4.9506421266334475e-05,
      "loss": 0.0321,
      "step": 7843,
      "task_loss": 0.053807858377695084
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999671924860143,
      "compression_loss": 0.0,
      "distillation_loss": 0.027329690754413605,
      "epoch": 7.45,
      "learning_rate": 4.9504312196213596e-05,
      "loss": 0.0253,
      "step": 7844,
      "task_loss": 0.007107492536306381
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999673615965774,
      "compression_loss": 0.0,
      "distillation_loss": 0.04351479932665825,
      "epoch": 7.45,
      "learning_rate": 4.95021986747664e-05,
      "loss": 0.0472,
      "step": 7845,
      "task_loss": 0.08041299134492874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799967530125005,
      "compression_loss": 0.0,
      "distillation_loss": 0.07271379977464676,
      "epoch": 7.45,
      "learning_rate": 4.9500080702376805e-05,
      "loss": 0.0735,
      "step": 7846,
      "task_loss": 0.08034893125295639
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999676980723006,
      "compression_loss": 0.0,
      "distillation_loss": 0.07482333481311798,
      "epoch": 7.45,
      "learning_rate": 4.949795827942956e-05,
      "loss": 0.087,
      "step": 7847,
      "task_loss": 0.19708333909511566
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799967865439468,
      "compression_loss": 0.0,
      "distillation_loss": 0.1008092612028122,
      "epoch": 7.45,
      "learning_rate": 4.9495831406310205e-05,
      "loss": 0.1136,
      "step": 7848,
      "task_loss": 0.22852961719036102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999680322275107,
      "compression_loss": 0.0,
      "distillation_loss": 0.05679786577820778,
      "epoch": 7.45,
      "learning_rate": 4.94937000834051e-05,
      "loss": 0.0553,
      "step": 7849,
      "task_loss": 0.042113713920116425
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999681984374326,
      "compression_loss": 0.0,
      "distillation_loss": 0.15729619562625885,
      "epoch": 7.45,
      "learning_rate": 4.9491564311101426e-05,
      "loss": 0.1526,
      "step": 7850,
      "task_loss": 0.1107601523399353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999683640702373,
      "compression_loss": 0.0,
      "distillation_loss": 0.0956711545586586,
      "epoch": 7.46,
      "learning_rate": 4.9489424089787125e-05,
      "loss": 0.0906,
      "step": 7851,
      "task_loss": 0.04521763324737549
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999685291269284,
      "compression_loss": 0.0,
      "distillation_loss": 0.1287081092596054,
      "epoch": 7.46,
      "learning_rate": 4.948727941985101e-05,
      "loss": 0.1281,
      "step": 7852,
      "task_loss": 0.1223960816860199
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999686936085098,
      "compression_loss": 0.0,
      "distillation_loss": 0.02583775669336319,
      "epoch": 7.46,
      "learning_rate": 4.948513030168265e-05,
      "loss": 0.0244,
      "step": 7853,
      "task_loss": 0.011274663731455803
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999688575159849,
      "compression_loss": 0.0,
      "distillation_loss": 0.016532419249415398,
      "epoch": 7.46,
      "learning_rate": 4.948297673567245e-05,
      "loss": 0.0152,
      "step": 7854,
      "task_loss": 0.0036627184599637985
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999690208503576,
      "compression_loss": 0.0,
      "distillation_loss": 0.11579355597496033,
      "epoch": 7.46,
      "learning_rate": 4.948081872221161e-05,
      "loss": 0.1192,
      "step": 7855,
      "task_loss": 0.1501166969537735
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999691836126315,
      "compression_loss": 0.0,
      "distillation_loss": 0.038752175867557526,
      "epoch": 7.46,
      "learning_rate": 4.9478656261692155e-05,
      "loss": 0.0355,
      "step": 7856,
      "task_loss": 0.0065444111824035645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999693458038103,
      "compression_loss": 0.0,
      "distillation_loss": 0.027601994574069977,
      "epoch": 7.46,
      "learning_rate": 4.947648935450689e-05,
      "loss": 0.0254,
      "step": 7857,
      "task_loss": 0.00563957542181015
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999695074248976,
      "compression_loss": 0.0,
      "distillation_loss": 0.10853047668933868,
      "epoch": 7.46,
      "learning_rate": 4.947431800104947e-05,
      "loss": 0.1053,
      "step": 7858,
      "task_loss": 0.07578997313976288
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999696684768972,
      "compression_loss": 0.0,
      "distillation_loss": 0.05468723922967911,
      "epoch": 7.46,
      "learning_rate": 4.94721422017143e-05,
      "loss": 0.0603,
      "step": 7859,
      "task_loss": 0.11080426722764969
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999698289608128,
      "compression_loss": 0.0,
      "distillation_loss": 0.05597386136651039,
      "epoch": 7.46,
      "learning_rate": 4.946996195689665e-05,
      "loss": 0.0563,
      "step": 7860,
      "task_loss": 0.0596843883395195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999699888776479,
      "compression_loss": 0.0,
      "distillation_loss": 0.0576602928340435,
      "epoch": 7.47,
      "learning_rate": 4.9467777266992555e-05,
      "loss": 0.0558,
      "step": 7861,
      "task_loss": 0.038872990757226944
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999701482284064,
      "compression_loss": 0.0,
      "distillation_loss": 0.032028764486312866,
      "epoch": 7.47,
      "learning_rate": 4.946558813239888e-05,
      "loss": 0.0351,
      "step": 7862,
      "task_loss": 0.06291679292917252
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999703070140919,
      "compression_loss": 0.0,
      "distillation_loss": 0.07386447489261627,
      "epoch": 7.47,
      "learning_rate": 4.94633945535133e-05,
      "loss": 0.071,
      "step": 7863,
      "task_loss": 0.04504578933119774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999704652357079,
      "compression_loss": 0.0,
      "distillation_loss": 0.026677310466766357,
      "epoch": 7.47,
      "learning_rate": 4.946119653073428e-05,
      "loss": 0.0249,
      "step": 7864,
      "task_loss": 0.009069759398698807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999706228942585,
      "compression_loss": 0.0,
      "distillation_loss": 0.025820819661021233,
      "epoch": 7.47,
      "learning_rate": 4.9458994064461103e-05,
      "loss": 0.0246,
      "step": 7865,
      "task_loss": 0.013258153572678566
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799970779990747,
      "compression_loss": 0.0,
      "distillation_loss": 0.15189041197299957,
      "epoch": 7.47,
      "learning_rate": 4.945678715509386e-05,
      "loss": 0.1558,
      "step": 7866,
      "task_loss": 0.1908566802740097
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999709365261771,
      "compression_loss": 0.0,
      "distillation_loss": 0.08925406634807587,
      "epoch": 7.47,
      "learning_rate": 4.9454575803033445e-05,
      "loss": 0.0941,
      "step": 7867,
      "task_loss": 0.13728086650371552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999710925015526,
      "compression_loss": 0.0,
      "distillation_loss": 0.12610575556755066,
      "epoch": 7.47,
      "learning_rate": 4.945236000868156e-05,
      "loss": 0.12,
      "step": 7868,
      "task_loss": 0.06535394489765167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999712479178773,
      "compression_loss": 0.0,
      "distillation_loss": 0.02531656250357628,
      "epoch": 7.47,
      "learning_rate": 4.9450139772440715e-05,
      "loss": 0.0236,
      "step": 7869,
      "task_loss": 0.007954435423016548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999714027761547,
      "compression_loss": 0.0,
      "distillation_loss": 0.02278076484799385,
      "epoch": 7.47,
      "learning_rate": 4.944791509471423e-05,
      "loss": 0.0209,
      "step": 7870,
      "task_loss": 0.0036298464983701706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999715570773887,
      "compression_loss": 0.0,
      "distillation_loss": 0.010984521359205246,
      "epoch": 7.47,
      "learning_rate": 4.944568597590622e-05,
      "loss": 0.0103,
      "step": 7871,
      "task_loss": 0.003945378586649895
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999717108225826,
      "compression_loss": 0.0,
      "distillation_loss": 0.011236527003347874,
      "epoch": 7.48,
      "learning_rate": 4.944345241642162e-05,
      "loss": 0.0105,
      "step": 7872,
      "task_loss": 0.0039033014327287674
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999718640127403,
      "compression_loss": 0.0,
      "distillation_loss": 0.11728902161121368,
      "epoch": 7.48,
      "learning_rate": 4.944121441666617e-05,
      "loss": 0.1264,
      "step": 7873,
      "task_loss": 0.2082228660583496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999720166488656,
      "compression_loss": 0.0,
      "distillation_loss": 0.05443963408470154,
      "epoch": 7.48,
      "learning_rate": 4.943897197704642e-05,
      "loss": 0.0619,
      "step": 7874,
      "task_loss": 0.12860585749149323
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999721687319621,
      "compression_loss": 0.0,
      "distillation_loss": 0.10132479667663574,
      "epoch": 7.48,
      "learning_rate": 4.9436725097969696e-05,
      "loss": 0.1139,
      "step": 7875,
      "task_loss": 0.22712013125419617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999723202630333,
      "compression_loss": 0.0,
      "distillation_loss": 0.020510438829660416,
      "epoch": 7.48,
      "learning_rate": 4.943447377984418e-05,
      "loss": 0.0188,
      "step": 7876,
      "task_loss": 0.0038789715617895126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999724712430831,
      "compression_loss": 0.0,
      "distillation_loss": 0.17946818470954895,
      "epoch": 7.48,
      "learning_rate": 4.943221802307882e-05,
      "loss": 0.1833,
      "step": 7877,
      "task_loss": 0.21781717240810394
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999726216731151,
      "compression_loss": 0.0,
      "distillation_loss": 0.050077952444553375,
      "epoch": 7.48,
      "learning_rate": 4.942995782808339e-05,
      "loss": 0.0547,
      "step": 7878,
      "task_loss": 0.09608137607574463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799972771554133,
      "compression_loss": 0.0,
      "distillation_loss": 0.03088299185037613,
      "epoch": 7.48,
      "learning_rate": 4.9427693195268466e-05,
      "loss": 0.0283,
      "step": 7879,
      "task_loss": 0.005498785525560379
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999729208871406,
      "compression_loss": 0.0,
      "distillation_loss": 0.0309526976197958,
      "epoch": 7.48,
      "learning_rate": 4.942542412504543e-05,
      "loss": 0.0317,
      "step": 7880,
      "task_loss": 0.038051947951316833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999730696731413,
      "compression_loss": 0.0,
      "distillation_loss": 0.028046930208802223,
      "epoch": 7.48,
      "learning_rate": 4.942315061782646e-05,
      "loss": 0.0358,
      "step": 7881,
      "task_loss": 0.10542275756597519
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799973217913139,
      "compression_loss": 0.0,
      "distillation_loss": 0.026341721415519714,
      "epoch": 7.49,
      "learning_rate": 4.942087267402457e-05,
      "loss": 0.0245,
      "step": 7882,
      "task_loss": 0.007860302925109863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999733656081373,
      "compression_loss": 0.0,
      "distillation_loss": 0.014681628905236721,
      "epoch": 7.49,
      "learning_rate": 4.941859029405353e-05,
      "loss": 0.0138,
      "step": 7883,
      "task_loss": 0.0060966480523347855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79997351275914,
      "compression_loss": 0.0,
      "distillation_loss": 0.17591474950313568,
      "epoch": 7.49,
      "learning_rate": 4.9416303478327974e-05,
      "loss": 0.1843,
      "step": 7884,
      "task_loss": 0.25975045561790466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999736593671505,
      "compression_loss": 0.0,
      "distillation_loss": 0.04054497182369232,
      "epoch": 7.49,
      "learning_rate": 4.9414012227263295e-05,
      "loss": 0.0438,
      "step": 7885,
      "task_loss": 0.07342273741960526
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999738054331729,
      "compression_loss": 0.0,
      "distillation_loss": 0.2347552627325058,
      "epoch": 7.49,
      "learning_rate": 4.941171654127572e-05,
      "loss": 0.2415,
      "step": 7886,
      "task_loss": 0.301849901676178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999739509582104,
      "compression_loss": 0.0,
      "distillation_loss": 0.04243939742445946,
      "epoch": 7.49,
      "learning_rate": 4.9409416420782264e-05,
      "loss": 0.0463,
      "step": 7887,
      "task_loss": 0.0810224711894989
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999740959432671,
      "compression_loss": 0.0,
      "distillation_loss": 0.021390702575445175,
      "epoch": 7.49,
      "learning_rate": 4.940711186620076e-05,
      "loss": 0.0198,
      "step": 7888,
      "task_loss": 0.0057023558765649796
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999742403893465,
      "compression_loss": 0.0,
      "distillation_loss": 0.08921066671609879,
      "epoch": 7.49,
      "learning_rate": 4.9404802877949843e-05,
      "loss": 0.1026,
      "step": 7889,
      "task_loss": 0.22274163365364075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999743842974523,
      "compression_loss": 0.0,
      "distillation_loss": 0.03275410085916519,
      "epoch": 7.49,
      "learning_rate": 4.940248945644894e-05,
      "loss": 0.0315,
      "step": 7890,
      "task_loss": 0.02053932100534439
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999745276685881,
      "compression_loss": 0.0,
      "distillation_loss": 0.11449338495731354,
      "epoch": 7.49,
      "learning_rate": 4.9400171602118306e-05,
      "loss": 0.1164,
      "step": 7891,
      "task_loss": 0.13319021463394165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999746705037577,
      "compression_loss": 0.0,
      "distillation_loss": 0.03880294784903526,
      "epoch": 7.49,
      "learning_rate": 4.939784931537899e-05,
      "loss": 0.0356,
      "step": 7892,
      "task_loss": 0.00670219212770462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999748128039648,
      "compression_loss": 0.0,
      "distillation_loss": 0.09690918028354645,
      "epoch": 7.5,
      "learning_rate": 4.9395522596652846e-05,
      "loss": 0.0992,
      "step": 7893,
      "task_loss": 0.12014244496822357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999749545702131,
      "compression_loss": 0.0,
      "distillation_loss": 0.025341130793094635,
      "epoch": 7.5,
      "learning_rate": 4.939319144636253e-05,
      "loss": 0.0233,
      "step": 7894,
      "task_loss": 0.005261138081550598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799975095803506,
      "compression_loss": 0.0,
      "distillation_loss": 0.0120600126683712,
      "epoch": 7.5,
      "learning_rate": 4.9390855864931504e-05,
      "loss": 0.0113,
      "step": 7895,
      "task_loss": 0.004162052646279335
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999752365048475,
      "compression_loss": 0.0,
      "distillation_loss": 0.1584700495004654,
      "epoch": 7.5,
      "learning_rate": 4.938851585278405e-05,
      "loss": 0.1588,
      "step": 7896,
      "task_loss": 0.16162604093551636
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999753766752412,
      "compression_loss": 0.0,
      "distillation_loss": 0.0646199956536293,
      "epoch": 7.5,
      "learning_rate": 4.938617141034523e-05,
      "loss": 0.0662,
      "step": 7897,
      "task_loss": 0.08005733042955399
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999755163156906,
      "compression_loss": 0.0,
      "distillation_loss": 0.08716416358947754,
      "epoch": 7.5,
      "learning_rate": 4.938382253804094e-05,
      "loss": 0.0858,
      "step": 7898,
      "task_loss": 0.07349246740341187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999756554271997,
      "compression_loss": 0.0,
      "distillation_loss": 0.022657334804534912,
      "epoch": 7.5,
      "learning_rate": 4.938146923629784e-05,
      "loss": 0.0267,
      "step": 7899,
      "task_loss": 0.06275828927755356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799975794010772,
      "compression_loss": 0.0,
      "distillation_loss": 0.01820671185851097,
      "epoch": 7.5,
      "learning_rate": 4.937911150554343e-05,
      "loss": 0.0228,
      "step": 7900,
      "task_loss": 0.06378181278705597
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999759320674112,
      "compression_loss": 0.0,
      "distillation_loss": 0.11242584884166718,
      "epoch": 7.5,
      "learning_rate": 4.9376749346206006e-05,
      "loss": 0.1087,
      "step": 7901,
      "task_loss": 0.07475525140762329
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799976069598121,
      "compression_loss": 0.0,
      "distillation_loss": 0.048631347715854645,
      "epoch": 7.5,
      "learning_rate": 4.937438275871467e-05,
      "loss": 0.058,
      "step": 7902,
      "task_loss": 0.1423451006412506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799976206603905,
      "compression_loss": 0.0,
      "distillation_loss": 0.035419534891843796,
      "epoch": 7.51,
      "learning_rate": 4.9372011743499315e-05,
      "loss": 0.0361,
      "step": 7903,
      "task_loss": 0.042088061571121216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999763430857669,
      "compression_loss": 0.0,
      "distillation_loss": 0.05880480259656906,
      "epoch": 7.51,
      "learning_rate": 4.9369636300990645e-05,
      "loss": 0.0785,
      "step": 7904,
      "task_loss": 0.2559299170970917
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999764790447106,
      "compression_loss": 0.0,
      "distillation_loss": 0.12047819793224335,
      "epoch": 7.51,
      "learning_rate": 4.936725643162018e-05,
      "loss": 0.1142,
      "step": 7905,
      "task_loss": 0.05795508623123169
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999766144817395,
      "compression_loss": 0.0,
      "distillation_loss": 0.11895313113927841,
      "epoch": 7.51,
      "learning_rate": 4.936487213582023e-05,
      "loss": 0.1289,
      "step": 7906,
      "task_loss": 0.2186514288187027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999767493978573,
      "compression_loss": 0.0,
      "distillation_loss": 0.08085181564092636,
      "epoch": 7.51,
      "learning_rate": 4.9362483414023905e-05,
      "loss": 0.0756,
      "step": 7907,
      "task_loss": 0.028206422924995422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799976883794068,
      "compression_loss": 0.0,
      "distillation_loss": 0.03390750288963318,
      "epoch": 7.51,
      "learning_rate": 4.936009026666515e-05,
      "loss": 0.0341,
      "step": 7908,
      "task_loss": 0.035543277859687805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999770176713749,
      "compression_loss": 0.0,
      "distillation_loss": 0.09985256940126419,
      "epoch": 7.51,
      "learning_rate": 4.935769269417867e-05,
      "loss": 0.1132,
      "step": 7909,
      "task_loss": 0.23357577621936798
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999771510307819,
      "compression_loss": 0.0,
      "distillation_loss": 0.14149194955825806,
      "epoch": 7.51,
      "learning_rate": 4.935529069700001e-05,
      "loss": 0.1658,
      "step": 7910,
      "task_loss": 0.3845874071121216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999772838732926,
      "compression_loss": 0.0,
      "distillation_loss": 0.12010196596384048,
      "epoch": 7.51,
      "learning_rate": 4.935288427556549e-05,
      "loss": 0.1175,
      "step": 7911,
      "task_loss": 0.0936579704284668
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999774161999107,
      "compression_loss": 0.0,
      "distillation_loss": 0.15679499506950378,
      "epoch": 7.51,
      "learning_rate": 4.935047343031227e-05,
      "loss": 0.1524,
      "step": 7912,
      "task_loss": 0.1123620867729187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999775480116399,
      "compression_loss": 0.0,
      "distillation_loss": 0.04103892296552658,
      "epoch": 7.51,
      "learning_rate": 4.934805816167827e-05,
      "loss": 0.0391,
      "step": 7913,
      "task_loss": 0.021613851189613342
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999776793094838,
      "compression_loss": 0.0,
      "distillation_loss": 0.0210304856300354,
      "epoch": 7.52,
      "learning_rate": 4.934563847010224e-05,
      "loss": 0.0286,
      "step": 7914,
      "task_loss": 0.09712889790534973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999778100944462,
      "compression_loss": 0.0,
      "distillation_loss": 0.049790140241384506,
      "epoch": 7.52,
      "learning_rate": 4.934321435602374e-05,
      "loss": 0.0514,
      "step": 7915,
      "task_loss": 0.06559078395366669
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999779403675308,
      "compression_loss": 0.0,
      "distillation_loss": 0.025263594463467598,
      "epoch": 7.52,
      "learning_rate": 4.934078581988311e-05,
      "loss": 0.0281,
      "step": 7916,
      "task_loss": 0.0537312813103199
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999780701297412,
      "compression_loss": 0.0,
      "distillation_loss": 0.03194242715835571,
      "epoch": 7.52,
      "learning_rate": 4.933835286212151e-05,
      "loss": 0.0295,
      "step": 7917,
      "task_loss": 0.007093427702784538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799978199382081,
      "compression_loss": 0.0,
      "distillation_loss": 0.056847527623176575,
      "epoch": 7.52,
      "learning_rate": 4.9335915483180896e-05,
      "loss": 0.0558,
      "step": 7918,
      "task_loss": 0.046506118029356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799978328125554,
      "compression_loss": 0.0,
      "distillation_loss": 0.03372214734554291,
      "epoch": 7.52,
      "learning_rate": 4.9333473683504025e-05,
      "loss": 0.0443,
      "step": 7919,
      "task_loss": 0.1395803838968277
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999784563611639,
      "compression_loss": 0.0,
      "distillation_loss": 0.08080513775348663,
      "epoch": 7.52,
      "learning_rate": 4.9331027463534484e-05,
      "loss": 0.0788,
      "step": 7920,
      "task_loss": 0.060925908386707306
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999785840899144,
      "compression_loss": 0.0,
      "distillation_loss": 0.018568016588687897,
      "epoch": 7.52,
      "learning_rate": 4.932857682371661e-05,
      "loss": 0.0259,
      "step": 7921,
      "task_loss": 0.09170715510845184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999787113128091,
      "compression_loss": 0.0,
      "distillation_loss": 0.03294629603624344,
      "epoch": 7.52,
      "learning_rate": 4.9326121764495596e-05,
      "loss": 0.0304,
      "step": 7922,
      "task_loss": 0.007955122739076614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999788380308517,
      "compression_loss": 0.0,
      "distillation_loss": 0.07836222648620605,
      "epoch": 7.52,
      "learning_rate": 4.932366228631741e-05,
      "loss": 0.0852,
      "step": 7923,
      "task_loss": 0.14636337757110596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999789642450458,
      "compression_loss": 0.0,
      "distillation_loss": 0.08306904882192612,
      "epoch": 7.53,
      "learning_rate": 4.932119838962882e-05,
      "loss": 0.0848,
      "step": 7924,
      "task_loss": 0.10019460320472717
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999790899563952,
      "compression_loss": 0.0,
      "distillation_loss": 0.02642223611474037,
      "epoch": 7.53,
      "learning_rate": 4.931873007487741e-05,
      "loss": 0.0414,
      "step": 7925,
      "task_loss": 0.17620962858200073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999792151659036,
      "compression_loss": 0.0,
      "distillation_loss": 0.030220147222280502,
      "epoch": 7.53,
      "learning_rate": 4.9316257342511565e-05,
      "loss": 0.0286,
      "step": 7926,
      "task_loss": 0.01419852301478386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999793398745746,
      "compression_loss": 0.0,
      "distillation_loss": 0.018758175894618034,
      "epoch": 7.53,
      "learning_rate": 4.9313780192980466e-05,
      "loss": 0.0236,
      "step": 7927,
      "task_loss": 0.06724384427070618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799979464083412,
      "compression_loss": 0.0,
      "distillation_loss": 0.14385569095611572,
      "epoch": 7.53,
      "learning_rate": 4.9311298626734095e-05,
      "loss": 0.1551,
      "step": 7928,
      "task_loss": 0.25620976090431213
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999795877934193,
      "compression_loss": 0.0,
      "distillation_loss": 0.13976441323757172,
      "epoch": 7.53,
      "learning_rate": 4.9308812644223245e-05,
      "loss": 0.1381,
      "step": 7929,
      "task_loss": 0.12334275990724564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999797110056004,
      "compression_loss": 0.0,
      "distillation_loss": 0.04338621348142624,
      "epoch": 7.53,
      "learning_rate": 4.9306322245899505e-05,
      "loss": 0.0628,
      "step": 7930,
      "task_loss": 0.23742617666721344
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999798337209587,
      "compression_loss": 0.0,
      "distillation_loss": 0.010634875856339931,
      "epoch": 7.53,
      "learning_rate": 4.930382743221528e-05,
      "loss": 0.0198,
      "step": 7931,
      "task_loss": 0.1026746854186058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999799559404982,
      "compression_loss": 0.0,
      "distillation_loss": 0.028556659817695618,
      "epoch": 7.53,
      "learning_rate": 4.930132820362374e-05,
      "loss": 0.0416,
      "step": 7932,
      "task_loss": 0.15901359915733337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999800776652223,
      "compression_loss": 0.0,
      "distillation_loss": 0.06336972862482071,
      "epoch": 7.53,
      "learning_rate": 4.9298824560578895e-05,
      "loss": 0.0625,
      "step": 7933,
      "task_loss": 0.054341040551662445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999801988961349,
      "compression_loss": 0.0,
      "distillation_loss": 0.09229055047035217,
      "epoch": 7.53,
      "learning_rate": 4.929631650353555e-05,
      "loss": 0.0899,
      "step": 7934,
      "task_loss": 0.06858476251363754
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999803196342395,
      "compression_loss": 0.0,
      "distillation_loss": 0.10068748891353607,
      "epoch": 7.54,
      "learning_rate": 4.92938040329493e-05,
      "loss": 0.0991,
      "step": 7935,
      "task_loss": 0.0846622884273529
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79998043988054,
      "compression_loss": 0.0,
      "distillation_loss": 0.033520594239234924,
      "epoch": 7.54,
      "learning_rate": 4.9291287149276544e-05,
      "loss": 0.0323,
      "step": 7936,
      "task_loss": 0.02134551852941513
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999805596360399,
      "compression_loss": 0.0,
      "distillation_loss": 0.04189582169055939,
      "epoch": 7.54,
      "learning_rate": 4.928876585297448e-05,
      "loss": 0.0512,
      "step": 7937,
      "task_loss": 0.13514596223831177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999806789017428,
      "compression_loss": 0.0,
      "distillation_loss": 0.04260486364364624,
      "epoch": 7.54,
      "learning_rate": 4.9286240144501136e-05,
      "loss": 0.0483,
      "step": 7938,
      "task_loss": 0.09960085153579712
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999807976786527,
      "compression_loss": 0.0,
      "distillation_loss": 0.08986024558544159,
      "epoch": 7.54,
      "learning_rate": 4.928371002431531e-05,
      "loss": 0.0864,
      "step": 7939,
      "task_loss": 0.05546959117054939
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999809159677731,
      "compression_loss": 0.0,
      "distillation_loss": 0.05223695933818817,
      "epoch": 7.54,
      "learning_rate": 4.92811754928766e-05,
      "loss": 0.0485,
      "step": 7940,
      "task_loss": 0.014858538284897804
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999810337701077,
      "compression_loss": 0.0,
      "distillation_loss": 0.022646090015769005,
      "epoch": 7.54,
      "learning_rate": 4.927863655064542e-05,
      "loss": 0.0294,
      "step": 7941,
      "task_loss": 0.0898606926202774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79998115108666,
      "compression_loss": 0.0,
      "distillation_loss": 0.12153077125549316,
      "epoch": 7.54,
      "learning_rate": 4.9276093198082986e-05,
      "loss": 0.1213,
      "step": 7942,
      "task_loss": 0.11907624453306198
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799981267918434,
      "compression_loss": 0.0,
      "distillation_loss": 0.10180039703845978,
      "epoch": 7.54,
      "learning_rate": 4.92735454356513e-05,
      "loss": 0.1072,
      "step": 7943,
      "task_loss": 0.15552875399589539
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999813842664332,
      "compression_loss": 0.0,
      "distillation_loss": 0.04333251342177391,
      "epoch": 7.54,
      "learning_rate": 4.927099326381319e-05,
      "loss": 0.0482,
      "step": 7944,
      "task_loss": 0.09174293279647827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999815001316612,
      "compression_loss": 0.0,
      "distillation_loss": 0.06807390600442886,
      "epoch": 7.55,
      "learning_rate": 4.926843668303227e-05,
      "loss": 0.0658,
      "step": 7945,
      "task_loss": 0.045711807906627655
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799981615515122,
      "compression_loss": 0.0,
      "distillation_loss": 0.0737372487783432,
      "epoch": 7.55,
      "learning_rate": 4.926587569377293e-05,
      "loss": 0.0728,
      "step": 7946,
      "task_loss": 0.06481769680976868
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799981730417819,
      "compression_loss": 0.0,
      "distillation_loss": 0.028609251603484154,
      "epoch": 7.55,
      "learning_rate": 4.926331029650042e-05,
      "loss": 0.0288,
      "step": 7947,
      "task_loss": 0.03084813803434372
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999818448407559,
      "compression_loss": 0.0,
      "distillation_loss": 0.10599102079868317,
      "epoch": 7.55,
      "learning_rate": 4.926074049168074e-05,
      "loss": 0.1118,
      "step": 7948,
      "task_loss": 0.16414104402065277
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999819587849365,
      "compression_loss": 0.0,
      "distillation_loss": 0.04272402077913284,
      "epoch": 7.55,
      "learning_rate": 4.9258166279780704e-05,
      "loss": 0.0482,
      "step": 7949,
      "task_loss": 0.0972161665558815
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999820722513644,
      "compression_loss": 0.0,
      "distillation_loss": 0.09052185714244843,
      "epoch": 7.55,
      "learning_rate": 4.925558766126794e-05,
      "loss": 0.0844,
      "step": 7950,
      "task_loss": 0.029212266206741333
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999821852410433,
      "compression_loss": 0.0,
      "distillation_loss": 0.031549785286188126,
      "epoch": 7.55,
      "learning_rate": 4.9253004636610856e-05,
      "loss": 0.0294,
      "step": 7951,
      "task_loss": 0.010255459696054459
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999822977549769,
      "compression_loss": 0.0,
      "distillation_loss": 0.02592216432094574,
      "epoch": 7.55,
      "learning_rate": 4.925041720627868e-05,
      "loss": 0.0242,
      "step": 7952,
      "task_loss": 0.008722037076950073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999824097941688,
      "compression_loss": 0.0,
      "distillation_loss": 0.051010869443416595,
      "epoch": 7.55,
      "learning_rate": 4.9247825370741416e-05,
      "loss": 0.0672,
      "step": 7953,
      "task_loss": 0.2124231904745102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999825213596229,
      "compression_loss": 0.0,
      "distillation_loss": 0.04993243142962456,
      "epoch": 7.55,
      "learning_rate": 4.924522913046991e-05,
      "loss": 0.0522,
      "step": 7954,
      "task_loss": 0.072673000395298
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999826324523426,
      "compression_loss": 0.0,
      "distillation_loss": 0.024953220039606094,
      "epoch": 7.55,
      "learning_rate": 4.924262848593576e-05,
      "loss": 0.0243,
      "step": 7955,
      "task_loss": 0.018649937584996223
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999827430733317,
      "compression_loss": 0.0,
      "distillation_loss": 0.07212147861719131,
      "epoch": 7.56,
      "learning_rate": 4.924002343761139e-05,
      "loss": 0.0838,
      "step": 7956,
      "task_loss": 0.18851712346076965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799982853223594,
      "compression_loss": 0.0,
      "distillation_loss": 0.012818554416298866,
      "epoch": 7.56,
      "learning_rate": 4.923741398597002e-05,
      "loss": 0.0167,
      "step": 7957,
      "task_loss": 0.051808781921863556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999829629041331,
      "compression_loss": 0.0,
      "distillation_loss": 0.12450156360864639,
      "epoch": 7.56,
      "learning_rate": 4.9234800131485675e-05,
      "loss": 0.1184,
      "step": 7958,
      "task_loss": 0.06334085762500763
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999830721159525,
      "compression_loss": 0.0,
      "distillation_loss": 0.0640573650598526,
      "epoch": 7.56,
      "learning_rate": 4.9232181874633164e-05,
      "loss": 0.0677,
      "step": 7959,
      "task_loss": 0.10021346807479858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999831808600562,
      "compression_loss": 0.0,
      "distillation_loss": 0.038610100746154785,
      "epoch": 7.56,
      "learning_rate": 4.922955921588812e-05,
      "loss": 0.0405,
      "step": 7960,
      "task_loss": 0.05722128972411156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999832891374477,
      "compression_loss": 0.0,
      "distillation_loss": 0.07282688468694687,
      "epoch": 7.56,
      "learning_rate": 4.922693215572695e-05,
      "loss": 0.0788,
      "step": 7961,
      "task_loss": 0.13281500339508057
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999833969491307,
      "compression_loss": 0.0,
      "distillation_loss": 0.04970156401395798,
      "epoch": 7.56,
      "learning_rate": 4.922430069462688e-05,
      "loss": 0.0521,
      "step": 7962,
      "task_loss": 0.07329516112804413
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999835042961089,
      "compression_loss": 0.0,
      "distillation_loss": 0.02577708661556244,
      "epoch": 7.56,
      "learning_rate": 4.9221664833065914e-05,
      "loss": 0.025,
      "step": 7963,
      "task_loss": 0.017895691096782684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999836111793859,
      "compression_loss": 0.0,
      "distillation_loss": 0.027007538825273514,
      "epoch": 7.56,
      "learning_rate": 4.921902457152289e-05,
      "loss": 0.0247,
      "step": 7964,
      "task_loss": 0.003965174779295921
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999837175999654,
      "compression_loss": 0.0,
      "distillation_loss": 0.019705362617969513,
      "epoch": 7.56,
      "learning_rate": 4.9216379910477403e-05,
      "loss": 0.0182,
      "step": 7965,
      "task_loss": 0.004658494144678116
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999838235588513,
      "compression_loss": 0.0,
      "distillation_loss": 0.033040136098861694,
      "epoch": 7.57,
      "learning_rate": 4.921373085040988e-05,
      "loss": 0.0342,
      "step": 7966,
      "task_loss": 0.04470521956682205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999839290570471,
      "compression_loss": 0.0,
      "distillation_loss": 0.023144233971834183,
      "epoch": 7.57,
      "learning_rate": 4.921107739180153e-05,
      "loss": 0.0267,
      "step": 7967,
      "task_loss": 0.05902532488107681
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999840340955565,
      "compression_loss": 0.0,
      "distillation_loss": 0.017418548464775085,
      "epoch": 7.57,
      "learning_rate": 4.9208419535134376e-05,
      "loss": 0.0189,
      "step": 7968,
      "task_loss": 0.03223051875829697
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999841386753831,
      "compression_loss": 0.0,
      "distillation_loss": 0.2720472514629364,
      "epoch": 7.57,
      "learning_rate": 4.920575728089122e-05,
      "loss": 0.2749,
      "step": 7969,
      "task_loss": 0.3010197579860687
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999842427975308,
      "compression_loss": 0.0,
      "distillation_loss": 0.052931733429431915,
      "epoch": 7.57,
      "learning_rate": 4.920309062955568e-05,
      "loss": 0.0564,
      "step": 7970,
      "task_loss": 0.0876016914844513
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799984346463003,
      "compression_loss": 0.0,
      "distillation_loss": 0.03811418265104294,
      "epoch": 7.57,
      "learning_rate": 4.920041958161217e-05,
      "loss": 0.0421,
      "step": 7971,
      "task_loss": 0.0776088535785675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999844496728036,
      "compression_loss": 0.0,
      "distillation_loss": 0.033870723098516464,
      "epoch": 7.57,
      "learning_rate": 4.9197744137545884e-05,
      "loss": 0.0475,
      "step": 7972,
      "task_loss": 0.1701067090034485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999845524279362,
      "compression_loss": 0.0,
      "distillation_loss": 0.02466290071606636,
      "epoch": 7.57,
      "learning_rate": 4.919506429784284e-05,
      "loss": 0.0317,
      "step": 7973,
      "task_loss": 0.09482477605342865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999846547294045,
      "compression_loss": 0.0,
      "distillation_loss": 0.053808264434337616,
      "epoch": 7.57,
      "learning_rate": 4.919238006298984e-05,
      "loss": 0.0581,
      "step": 7974,
      "task_loss": 0.0962759405374527
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999847565782121,
      "compression_loss": 0.0,
      "distillation_loss": 0.051727067679166794,
      "epoch": 7.57,
      "learning_rate": 4.9189691433474494e-05,
      "loss": 0.064,
      "step": 7975,
      "task_loss": 0.17400768399238586
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999848579753629,
      "compression_loss": 0.0,
      "distillation_loss": 0.029435984790325165,
      "epoch": 7.57,
      "learning_rate": 4.91869984097852e-05,
      "loss": 0.0384,
      "step": 7976,
      "task_loss": 0.11923931539058685
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999849589218603,
      "compression_loss": 0.0,
      "distillation_loss": 0.030128782615065575,
      "epoch": 7.58,
      "learning_rate": 4.918430099241116e-05,
      "loss": 0.036,
      "step": 7977,
      "task_loss": 0.08850212395191193
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999850594187082,
      "compression_loss": 0.0,
      "distillation_loss": 0.03268412873148918,
      "epoch": 7.58,
      "learning_rate": 4.918159918184236e-05,
      "loss": 0.0301,
      "step": 7978,
      "task_loss": 0.006716296076774597
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999851594669102,
      "compression_loss": 0.0,
      "distillation_loss": 0.03931747376918793,
      "epoch": 7.58,
      "learning_rate": 4.9178892978569625e-05,
      "loss": 0.0459,
      "step": 7979,
      "task_loss": 0.10497306287288666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79998525906747,
      "compression_loss": 0.0,
      "distillation_loss": 0.015478894114494324,
      "epoch": 7.58,
      "learning_rate": 4.9176182383084524e-05,
      "loss": 0.0297,
      "step": 7980,
      "task_loss": 0.15772272646427155
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999853582213913,
      "compression_loss": 0.0,
      "distillation_loss": 0.09228493273258209,
      "epoch": 7.58,
      "learning_rate": 4.917346739587946e-05,
      "loss": 0.0927,
      "step": 7981,
      "task_loss": 0.09685853123664856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999854569296777,
      "compression_loss": 0.0,
      "distillation_loss": 0.01883935183286667,
      "epoch": 7.58,
      "learning_rate": 4.917074801744763e-05,
      "loss": 0.0178,
      "step": 7982,
      "task_loss": 0.00816972553730011
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999855551933329,
      "compression_loss": 0.0,
      "distillation_loss": 0.021920818835496902,
      "epoch": 7.58,
      "learning_rate": 4.916802424828301e-05,
      "loss": 0.0203,
      "step": 7983,
      "task_loss": 0.0058016423135995865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999856530133607,
      "compression_loss": 0.0,
      "distillation_loss": 0.02121850848197937,
      "epoch": 7.58,
      "learning_rate": 4.9165296088880384e-05,
      "loss": 0.0198,
      "step": 7984,
      "task_loss": 0.0068349651992321014
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999857503907646,
      "compression_loss": 0.0,
      "distillation_loss": 0.11084365099668503,
      "epoch": 7.58,
      "learning_rate": 4.916256353973535e-05,
      "loss": 0.1116,
      "step": 7985,
      "task_loss": 0.11807304620742798
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999858473265484,
      "compression_loss": 0.0,
      "distillation_loss": 0.031100111082196236,
      "epoch": 7.58,
      "learning_rate": 4.9159826601344286e-05,
      "loss": 0.0335,
      "step": 7986,
      "task_loss": 0.05468904972076416
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999859438217158,
      "compression_loss": 0.0,
      "distillation_loss": 0.058571815490722656,
      "epoch": 7.58,
      "learning_rate": 4.915708527420435e-05,
      "loss": 0.065,
      "step": 7987,
      "task_loss": 0.1225350946187973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999860398772705,
      "compression_loss": 0.0,
      "distillation_loss": 0.033145397901535034,
      "epoch": 7.59,
      "learning_rate": 4.9154339558813546e-05,
      "loss": 0.0307,
      "step": 7988,
      "task_loss": 0.008651839569211006
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999861354942159,
      "compression_loss": 0.0,
      "distillation_loss": 0.11876051872968674,
      "epoch": 7.59,
      "learning_rate": 4.915158945567062e-05,
      "loss": 0.1123,
      "step": 7989,
      "task_loss": 0.05439896881580353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999862306735561,
      "compression_loss": 0.0,
      "distillation_loss": 0.1755588948726654,
      "epoch": 7.59,
      "learning_rate": 4.914883496527516e-05,
      "loss": 0.1739,
      "step": 7990,
      "task_loss": 0.15885521471500397
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999863254162946,
      "compression_loss": 0.0,
      "distillation_loss": 0.06931862980127335,
      "epoch": 7.59,
      "learning_rate": 4.914607608812753e-05,
      "loss": 0.0661,
      "step": 7991,
      "task_loss": 0.03663593530654907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799986419723435,
      "compression_loss": 0.0,
      "distillation_loss": 0.0520208366215229,
      "epoch": 7.59,
      "learning_rate": 4.9143312824728896e-05,
      "loss": 0.0687,
      "step": 7992,
      "task_loss": 0.2192659229040146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799986513595981,
      "compression_loss": 0.0,
      "distillation_loss": 0.045383553951978683,
      "epoch": 7.59,
      "learning_rate": 4.91405451755812e-05,
      "loss": 0.0435,
      "step": 7993,
      "task_loss": 0.026806168258190155
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999866070349365,
      "compression_loss": 0.0,
      "distillation_loss": 0.013009166345000267,
      "epoch": 7.59,
      "learning_rate": 4.913777314118721e-05,
      "loss": 0.0176,
      "step": 7994,
      "task_loss": 0.05919176712632179
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999867000413049,
      "compression_loss": 0.0,
      "distillation_loss": 0.0388064906001091,
      "epoch": 7.59,
      "learning_rate": 4.9134996722050483e-05,
      "loss": 0.0513,
      "step": 7995,
      "task_loss": 0.16394969820976257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999867926160901,
      "compression_loss": 0.0,
      "distillation_loss": 0.018643934279680252,
      "epoch": 7.59,
      "learning_rate": 4.913221591867537e-05,
      "loss": 0.0346,
      "step": 7996,
      "task_loss": 0.17829307913780212
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999868847602956,
      "compression_loss": 0.0,
      "distillation_loss": 0.013243372552096844,
      "epoch": 7.59,
      "learning_rate": 4.912943073156701e-05,
      "loss": 0.0122,
      "step": 7997,
      "task_loss": 0.0027358736842870712
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999869764749251,
      "compression_loss": 0.0,
      "distillation_loss": 0.036573659628629684,
      "epoch": 7.6,
      "learning_rate": 4.912664116123134e-05,
      "loss": 0.0333,
      "step": 7998,
      "task_loss": 0.003510754555463791
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999870677609825,
      "compression_loss": 0.0,
      "distillation_loss": 0.03176216036081314,
      "epoch": 7.6,
      "learning_rate": 4.9123847208175126e-05,
      "loss": 0.0383,
      "step": 7999,
      "task_loss": 0.09740820527076721
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999871586194712,
      "compression_loss": 0.0,
      "distillation_loss": 0.013369059190154076,
      "epoch": 7.6,
      "learning_rate": 4.912104887290587e-05,
      "loss": 0.0123,
      "step": 8000,
      "task_loss": 0.0031455066055059433
    },
    {
      "epoch": 7.6,
      "eval_accuracy": 0.8784403669724771,
      "eval_loss": 0.49849244952201843,
      "eval_runtime": 18.0456,
      "eval_samples_per_second": 48.322,
      "eval_steps_per_second": 6.04,
      "step": 8000
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999872490513951,
      "compression_loss": 0.0,
      "distillation_loss": 0.10992217063903809,
      "epoch": 7.6,
      "learning_rate": 4.911824615593193e-05,
      "loss": 0.1131,
      "step": 8001,
      "task_loss": 0.1418878138065338
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999873390577578,
      "compression_loss": 0.0,
      "distillation_loss": 0.042782656848430634,
      "epoch": 7.6,
      "learning_rate": 4.9115439057762416e-05,
      "loss": 0.0481,
      "step": 8002,
      "task_loss": 0.09572674334049225
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999874286395628,
      "compression_loss": 0.0,
      "distillation_loss": 0.03839121758937836,
      "epoch": 7.6,
      "learning_rate": 4.911262757890726e-05,
      "loss": 0.0471,
      "step": 8003,
      "task_loss": 0.12578138709068298
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999875177978142,
      "compression_loss": 0.0,
      "distillation_loss": 0.01663174107670784,
      "epoch": 7.6,
      "learning_rate": 4.9109811719877166e-05,
      "loss": 0.024,
      "step": 8004,
      "task_loss": 0.09052921831607819
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999876065335153,
      "compression_loss": 0.0,
      "distillation_loss": 0.09784665703773499,
      "epoch": 7.6,
      "learning_rate": 4.910699148118367e-05,
      "loss": 0.0997,
      "step": 8005,
      "task_loss": 0.1160675436258316
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79998769484767,
      "compression_loss": 0.0,
      "distillation_loss": 0.027502398937940598,
      "epoch": 7.6,
      "learning_rate": 4.910416686333906e-05,
      "loss": 0.0349,
      "step": 8006,
      "task_loss": 0.10173024237155914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999877827412818,
      "compression_loss": 0.0,
      "distillation_loss": 0.03942710533738136,
      "epoch": 7.6,
      "learning_rate": 4.910133786685646e-05,
      "loss": 0.0418,
      "step": 8007,
      "task_loss": 0.06354475021362305
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999878702153546,
      "compression_loss": 0.0,
      "distillation_loss": 0.016465935856103897,
      "epoch": 7.6,
      "learning_rate": 4.9098504492249764e-05,
      "loss": 0.0239,
      "step": 8008,
      "task_loss": 0.09078645706176758
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999879572708918,
      "compression_loss": 0.0,
      "distillation_loss": 0.020133644342422485,
      "epoch": 7.61,
      "learning_rate": 4.9095666740033664e-05,
      "loss": 0.0187,
      "step": 8009,
      "task_loss": 0.006201488897204399
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999880439088973,
      "compression_loss": 0.0,
      "distillation_loss": 0.04203331097960472,
      "epoch": 7.61,
      "learning_rate": 4.9092824610723655e-05,
      "loss": 0.0508,
      "step": 8010,
      "task_loss": 0.12994877994060516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999881301303748,
      "compression_loss": 0.0,
      "distillation_loss": 0.04908212646842003,
      "epoch": 7.61,
      "learning_rate": 4.908997810483602e-05,
      "loss": 0.0532,
      "step": 8011,
      "task_loss": 0.09064097702503204
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799988215936328,
      "compression_loss": 0.0,
      "distillation_loss": 0.0626966580748558,
      "epoch": 7.61,
      "learning_rate": 4.908712722288785e-05,
      "loss": 0.088,
      "step": 8012,
      "task_loss": 0.316189706325531
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999883013277603,
      "compression_loss": 0.0,
      "distillation_loss": 0.058064427226781845,
      "epoch": 7.61,
      "learning_rate": 4.9084271965397014e-05,
      "loss": 0.0626,
      "step": 8013,
      "task_loss": 0.1036590188741684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999883863056757,
      "compression_loss": 0.0,
      "distillation_loss": 0.04076586291193962,
      "epoch": 7.61,
      "learning_rate": 4.908141233288218e-05,
      "loss": 0.0374,
      "step": 8014,
      "task_loss": 0.00735941156744957
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999884708710777,
      "compression_loss": 0.0,
      "distillation_loss": 0.08777140825986862,
      "epoch": 7.61,
      "learning_rate": 4.907854832586282e-05,
      "loss": 0.0976,
      "step": 8015,
      "task_loss": 0.18588636815547943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999885550249701,
      "compression_loss": 0.0,
      "distillation_loss": 0.038717057555913925,
      "epoch": 7.61,
      "learning_rate": 4.907567994485919e-05,
      "loss": 0.041,
      "step": 8016,
      "task_loss": 0.06178871542215347
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999886387683565,
      "compression_loss": 0.0,
      "distillation_loss": 0.14275038242340088,
      "epoch": 7.61,
      "learning_rate": 4.9072807190392354e-05,
      "loss": 0.1415,
      "step": 8017,
      "task_loss": 0.13036131858825684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999887221022406,
      "compression_loss": 0.0,
      "distillation_loss": 0.024897240102291107,
      "epoch": 7.61,
      "learning_rate": 4.906993006298416e-05,
      "loss": 0.0451,
      "step": 8018,
      "task_loss": 0.22650375962257385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999888050276261,
      "compression_loss": 0.0,
      "distillation_loss": 0.14283829927444458,
      "epoch": 7.62,
      "learning_rate": 4.9067048563157235e-05,
      "loss": 0.1404,
      "step": 8019,
      "task_loss": 0.1185140609741211
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999888875455167,
      "compression_loss": 0.0,
      "distillation_loss": 0.08865389972925186,
      "epoch": 7.62,
      "learning_rate": 4.906416269143505e-05,
      "loss": 0.0981,
      "step": 8020,
      "task_loss": 0.18301840126514435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799988969656916,
      "compression_loss": 0.0,
      "distillation_loss": 0.08406002819538116,
      "epoch": 7.62,
      "learning_rate": 4.90612724483418e-05,
      "loss": 0.0969,
      "step": 8021,
      "task_loss": 0.21275919675827026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999890513628277,
      "compression_loss": 0.0,
      "distillation_loss": 0.08485761284828186,
      "epoch": 7.62,
      "learning_rate": 4.905837783440253e-05,
      "loss": 0.0828,
      "step": 8022,
      "task_loss": 0.0644686371088028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999891326642555,
      "compression_loss": 0.0,
      "distillation_loss": 0.040136683732271194,
      "epoch": 7.62,
      "learning_rate": 4.905547885014307e-05,
      "loss": 0.0373,
      "step": 8023,
      "task_loss": 0.01128750666975975
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999892135622032,
      "compression_loss": 0.0,
      "distillation_loss": 0.014311270788311958,
      "epoch": 7.62,
      "learning_rate": 4.9052575496090016e-05,
      "loss": 0.0133,
      "step": 8024,
      "task_loss": 0.004563674330711365
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999892940576744,
      "compression_loss": 0.0,
      "distillation_loss": 0.03336448222398758,
      "epoch": 7.62,
      "learning_rate": 4.904966777277079e-05,
      "loss": 0.0309,
      "step": 8025,
      "task_loss": 0.008388657122850418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999893741516727,
      "compression_loss": 0.0,
      "distillation_loss": 0.030747603625059128,
      "epoch": 7.62,
      "learning_rate": 4.9046755680713586e-05,
      "loss": 0.033,
      "step": 8026,
      "task_loss": 0.05296764522790909
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999894538452018,
      "compression_loss": 0.0,
      "distillation_loss": 0.021251916885375977,
      "epoch": 7.62,
      "learning_rate": 4.90438392204474e-05,
      "loss": 0.0355,
      "step": 8027,
      "task_loss": 0.16398029029369354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999895331392655,
      "compression_loss": 0.0,
      "distillation_loss": 0.21242570877075195,
      "epoch": 7.62,
      "learning_rate": 4.9040918392502026e-05,
      "loss": 0.2064,
      "step": 8028,
      "task_loss": 0.15173274278640747
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999896120348674,
      "compression_loss": 0.0,
      "distillation_loss": 0.020135043188929558,
      "epoch": 7.62,
      "learning_rate": 4.903799319740804e-05,
      "loss": 0.0209,
      "step": 8029,
      "task_loss": 0.02775605395436287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999896905330112,
      "compression_loss": 0.0,
      "distillation_loss": 0.033395953476428986,
      "epoch": 7.63,
      "learning_rate": 4.903506363569683e-05,
      "loss": 0.0337,
      "step": 8030,
      "task_loss": 0.036266930401325226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999897686347005,
      "compression_loss": 0.0,
      "distillation_loss": 0.02742244489490986,
      "epoch": 7.63,
      "learning_rate": 4.9032129707900556e-05,
      "loss": 0.0256,
      "step": 8031,
      "task_loss": 0.008756054565310478
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999898463409392,
      "compression_loss": 0.0,
      "distillation_loss": 0.023902807384729385,
      "epoch": 7.63,
      "learning_rate": 4.9029191414552165e-05,
      "loss": 0.0294,
      "step": 8032,
      "task_loss": 0.07919642329216003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999899236527307,
      "compression_loss": 0.0,
      "distillation_loss": 0.03757765516638756,
      "epoch": 7.63,
      "learning_rate": 4.9026248756185445e-05,
      "loss": 0.0356,
      "step": 8033,
      "task_loss": 0.017600098624825478
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999900005710788,
      "compression_loss": 0.0,
      "distillation_loss": 0.08396127820014954,
      "epoch": 7.63,
      "learning_rate": 4.902330173333492e-05,
      "loss": 0.0825,
      "step": 8034,
      "task_loss": 0.06926584243774414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999900770969873,
      "compression_loss": 0.0,
      "distillation_loss": 0.03239300101995468,
      "epoch": 7.63,
      "learning_rate": 4.9020350346535936e-05,
      "loss": 0.0354,
      "step": 8035,
      "task_loss": 0.062150366604328156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999901532314597,
      "compression_loss": 0.0,
      "distillation_loss": 0.0414559543132782,
      "epoch": 7.63,
      "learning_rate": 4.901739459632463e-05,
      "loss": 0.0448,
      "step": 8036,
      "task_loss": 0.07490754127502441
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999902289754999,
      "compression_loss": 0.0,
      "distillation_loss": 0.03927509859204292,
      "epoch": 7.63,
      "learning_rate": 4.901443448323792e-05,
      "loss": 0.0396,
      "step": 8037,
      "task_loss": 0.04217911139130592
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999903043301113,
      "compression_loss": 0.0,
      "distillation_loss": 0.10875845700502396,
      "epoch": 7.63,
      "learning_rate": 4.901147000781355e-05,
      "loss": 0.1031,
      "step": 8038,
      "task_loss": 0.05257695913314819
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999903792962978,
      "compression_loss": 0.0,
      "distillation_loss": 0.18646162748336792,
      "epoch": 7.63,
      "learning_rate": 4.9008501170589996e-05,
      "loss": 0.1765,
      "step": 8039,
      "task_loss": 0.08664773404598236
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799990453875063,
      "compression_loss": 0.0,
      "distillation_loss": 0.13723516464233398,
      "epoch": 7.64,
      "learning_rate": 4.900552797210658e-05,
      "loss": 0.1398,
      "step": 8040,
      "task_loss": 0.16254064440727234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999905280674106,
      "compression_loss": 0.0,
      "distillation_loss": 0.05691935867071152,
      "epoch": 7.64,
      "learning_rate": 4.90025504129034e-05,
      "loss": 0.0737,
      "step": 8041,
      "task_loss": 0.22483864426612854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999906018743442,
      "compression_loss": 0.0,
      "distillation_loss": 0.017723996192216873,
      "epoch": 7.64,
      "learning_rate": 4.8999568493521345e-05,
      "loss": 0.0164,
      "step": 8042,
      "task_loss": 0.004084032028913498
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999906752968677,
      "compression_loss": 0.0,
      "distillation_loss": 0.07552896440029144,
      "epoch": 7.64,
      "learning_rate": 4.899658221450208e-05,
      "loss": 0.0805,
      "step": 8043,
      "task_loss": 0.1257321536540985
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999907483359846,
      "compression_loss": 0.0,
      "distillation_loss": 0.10802210122346878,
      "epoch": 7.64,
      "learning_rate": 4.899359157638809e-05,
      "loss": 0.1099,
      "step": 8044,
      "task_loss": 0.12680913507938385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999908209926986,
      "compression_loss": 0.0,
      "distillation_loss": 0.12618671357631683,
      "epoch": 7.64,
      "learning_rate": 4.899059657972264e-05,
      "loss": 0.1189,
      "step": 8045,
      "task_loss": 0.0529848150908947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999908932680134,
      "compression_loss": 0.0,
      "distillation_loss": 0.027106711640954018,
      "epoch": 7.64,
      "learning_rate": 4.898759722504977e-05,
      "loss": 0.0254,
      "step": 8046,
      "task_loss": 0.010013708844780922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999909651629327,
      "compression_loss": 0.0,
      "distillation_loss": 0.02689509466290474,
      "epoch": 7.64,
      "learning_rate": 4.8984593512914356e-05,
      "loss": 0.0332,
      "step": 8047,
      "task_loss": 0.08948872238397598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999910366784602,
      "compression_loss": 0.0,
      "distillation_loss": 0.19355759024620056,
      "epoch": 7.64,
      "learning_rate": 4.898158544386201e-05,
      "loss": 0.1822,
      "step": 8048,
      "task_loss": 0.08036290854215622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999911078155995,
      "compression_loss": 0.0,
      "distillation_loss": 0.029629534110426903,
      "epoch": 7.64,
      "learning_rate": 4.897857301843917e-05,
      "loss": 0.0298,
      "step": 8049,
      "task_loss": 0.030841048806905746
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999911785753543,
      "compression_loss": 0.0,
      "distillation_loss": 0.12482485175132751,
      "epoch": 7.64,
      "learning_rate": 4.897555623719306e-05,
      "loss": 0.129,
      "step": 8050,
      "task_loss": 0.1669907122850418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999912489587283,
      "compression_loss": 0.0,
      "distillation_loss": 0.11259239912033081,
      "epoch": 7.65,
      "learning_rate": 4.897253510067169e-05,
      "loss": 0.1228,
      "step": 8051,
      "task_loss": 0.21496890485286713
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999913189667253,
      "compression_loss": 0.0,
      "distillation_loss": 0.10157469660043716,
      "epoch": 7.65,
      "learning_rate": 4.896950960942387e-05,
      "loss": 0.1013,
      "step": 8052,
      "task_loss": 0.09880008548498154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999913886003488,
      "compression_loss": 0.0,
      "distillation_loss": 0.06962237507104874,
      "epoch": 7.65,
      "learning_rate": 4.896647976399919e-05,
      "loss": 0.0682,
      "step": 8053,
      "task_loss": 0.05498020350933075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999914578606026,
      "compression_loss": 0.0,
      "distillation_loss": 0.02622218243777752,
      "epoch": 7.65,
      "learning_rate": 4.896344556494804e-05,
      "loss": 0.0248,
      "step": 8054,
      "task_loss": 0.011739548295736313
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999915267484903,
      "compression_loss": 0.0,
      "distillation_loss": 0.05761996656656265,
      "epoch": 7.65,
      "learning_rate": 4.8960407012821584e-05,
      "loss": 0.0619,
      "step": 8055,
      "task_loss": 0.10011887550354004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999915952650157,
      "compression_loss": 0.0,
      "distillation_loss": 0.028014160692691803,
      "epoch": 7.65,
      "learning_rate": 4.895736410817181e-05,
      "loss": 0.026,
      "step": 8056,
      "task_loss": 0.007695335894823074
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999916634111823,
      "compression_loss": 0.0,
      "distillation_loss": 0.018892180174589157,
      "epoch": 7.65,
      "learning_rate": 4.8954316851551465e-05,
      "loss": 0.0177,
      "step": 8057,
      "task_loss": 0.006759140640497208
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799991731187994,
      "compression_loss": 0.0,
      "distillation_loss": 0.13491389155387878,
      "epoch": 7.65,
      "learning_rate": 4.895126524351409e-05,
      "loss": 0.1503,
      "step": 8058,
      "task_loss": 0.2890867292881012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999917985964543,
      "compression_loss": 0.0,
      "distillation_loss": 0.04141715168952942,
      "epoch": 7.65,
      "learning_rate": 4.8948209284614046e-05,
      "loss": 0.039,
      "step": 8059,
      "task_loss": 0.01771230250597
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999918656375671,
      "compression_loss": 0.0,
      "distillation_loss": 0.016739333048462868,
      "epoch": 7.65,
      "learning_rate": 4.894514897540643e-05,
      "loss": 0.0158,
      "step": 8060,
      "task_loss": 0.007118521258234978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999919323123358,
      "compression_loss": 0.0,
      "distillation_loss": 0.01123537216335535,
      "epoch": 7.66,
      "learning_rate": 4.89420843164472e-05,
      "loss": 0.0202,
      "step": 8061,
      "task_loss": 0.10039521753787994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999919986217642,
      "compression_loss": 0.0,
      "distillation_loss": 0.019477343186736107,
      "epoch": 7.66,
      "learning_rate": 4.893901530829304e-05,
      "loss": 0.0179,
      "step": 8062,
      "task_loss": 0.0040080491453409195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799992064566856,
      "compression_loss": 0.0,
      "distillation_loss": 0.04232557862997055,
      "epoch": 7.66,
      "learning_rate": 4.8935941951501463e-05,
      "loss": 0.0474,
      "step": 8063,
      "task_loss": 0.0932355746626854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999921301486149,
      "compression_loss": 0.0,
      "distillation_loss": 0.10348470509052277,
      "epoch": 7.66,
      "learning_rate": 4.893286424663075e-05,
      "loss": 0.0977,
      "step": 8064,
      "task_loss": 0.045336902141571045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999921953680447,
      "compression_loss": 0.0,
      "distillation_loss": 0.01743854209780693,
      "epoch": 7.66,
      "learning_rate": 4.892978219423998e-05,
      "loss": 0.0331,
      "step": 8065,
      "task_loss": 0.17381621897220612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999922602261488,
      "compression_loss": 0.0,
      "distillation_loss": 0.01273958757519722,
      "epoch": 7.66,
      "learning_rate": 4.892669579488903e-05,
      "loss": 0.0186,
      "step": 8066,
      "task_loss": 0.07183681428432465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799992324723931,
      "compression_loss": 0.0,
      "distillation_loss": 0.11235454678535461,
      "epoch": 7.66,
      "learning_rate": 4.892360504913856e-05,
      "loss": 0.1105,
      "step": 8067,
      "task_loss": 0.09365284442901611
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999923888623951,
      "compression_loss": 0.0,
      "distillation_loss": 0.05907963961362839,
      "epoch": 7.66,
      "learning_rate": 4.8920509957550016e-05,
      "loss": 0.054,
      "step": 8068,
      "task_loss": 0.008163506165146828
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999924526425447,
      "compression_loss": 0.0,
      "distillation_loss": 0.023158248513936996,
      "epoch": 7.66,
      "learning_rate": 4.8917410520685635e-05,
      "loss": 0.0351,
      "step": 8069,
      "task_loss": 0.14244556427001953
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999925160653834,
      "compression_loss": 0.0,
      "distillation_loss": 0.016128631308674812,
      "epoch": 7.66,
      "learning_rate": 4.891430673910844e-05,
      "loss": 0.015,
      "step": 8070,
      "task_loss": 0.0045019593089818954
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799992579131915,
      "compression_loss": 0.0,
      "distillation_loss": 0.023808129131793976,
      "epoch": 7.66,
      "learning_rate": 4.891119861338226e-05,
      "loss": 0.022,
      "step": 8071,
      "task_loss": 0.005267852917313576
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999926418431431,
      "compression_loss": 0.0,
      "distillation_loss": 0.02601657807826996,
      "epoch": 7.67,
      "learning_rate": 4.8908086144071694e-05,
      "loss": 0.0241,
      "step": 8072,
      "task_loss": 0.0071242451667785645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999927042000715,
      "compression_loss": 0.0,
      "distillation_loss": 0.045584745705127716,
      "epoch": 7.67,
      "learning_rate": 4.8904969331742136e-05,
      "loss": 0.0496,
      "step": 8073,
      "task_loss": 0.08568625897169113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999927662037037,
      "compression_loss": 0.0,
      "distillation_loss": 0.021556910127401352,
      "epoch": 7.67,
      "learning_rate": 4.890184817695976e-05,
      "loss": 0.0281,
      "step": 8074,
      "task_loss": 0.08685538172721863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999928278550436,
      "compression_loss": 0.0,
      "distillation_loss": 0.09852443635463715,
      "epoch": 7.67,
      "learning_rate": 4.8898722680291564e-05,
      "loss": 0.1017,
      "step": 8075,
      "task_loss": 0.13034701347351074
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999928891550947,
      "compression_loss": 0.0,
      "distillation_loss": 0.034188371151685715,
      "epoch": 7.67,
      "learning_rate": 4.8895592842305295e-05,
      "loss": 0.0401,
      "step": 8076,
      "task_loss": 0.09356120228767395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999929501048607,
      "compression_loss": 0.0,
      "distillation_loss": 0.02725556679069996,
      "epoch": 7.67,
      "learning_rate": 4.88924586635695e-05,
      "loss": 0.0348,
      "step": 8077,
      "task_loss": 0.1023518443107605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999930107053455,
      "compression_loss": 0.0,
      "distillation_loss": 0.0407828688621521,
      "epoch": 7.67,
      "learning_rate": 4.888932014465352e-05,
      "loss": 0.0427,
      "step": 8078,
      "task_loss": 0.060073304921388626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999930709575525,
      "compression_loss": 0.0,
      "distillation_loss": 0.02005520462989807,
      "epoch": 7.67,
      "learning_rate": 4.888617728612749e-05,
      "loss": 0.0187,
      "step": 8079,
      "task_loss": 0.006268629804253578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999931308624855,
      "compression_loss": 0.0,
      "distillation_loss": 0.016719898208975792,
      "epoch": 7.67,
      "learning_rate": 4.888303008856231e-05,
      "loss": 0.0154,
      "step": 8080,
      "task_loss": 0.0033553019165992737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999931904211481,
      "compression_loss": 0.0,
      "distillation_loss": 0.01692567765712738,
      "epoch": 7.67,
      "learning_rate": 4.88798785525297e-05,
      "loss": 0.0213,
      "step": 8081,
      "task_loss": 0.060581427067518234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999932496345442,
      "compression_loss": 0.0,
      "distillation_loss": 0.03452851623296738,
      "epoch": 7.68,
      "learning_rate": 4.887672267860214e-05,
      "loss": 0.0371,
      "step": 8082,
      "task_loss": 0.059820882976055145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999933085036772,
      "compression_loss": 0.0,
      "distillation_loss": 0.022003114223480225,
      "epoch": 7.68,
      "learning_rate": 4.887356246735292e-05,
      "loss": 0.0203,
      "step": 8083,
      "task_loss": 0.004899036139249802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799993367029551,
      "compression_loss": 0.0,
      "distillation_loss": 0.07817333936691284,
      "epoch": 7.68,
      "learning_rate": 4.8870397919356094e-05,
      "loss": 0.0767,
      "step": 8084,
      "task_loss": 0.0634901374578476
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999934252131693,
      "compression_loss": 0.0,
      "distillation_loss": 0.08945944160223007,
      "epoch": 7.68,
      "learning_rate": 4.8867229035186526e-05,
      "loss": 0.0865,
      "step": 8085,
      "task_loss": 0.060247667133808136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999934830555356,
      "compression_loss": 0.0,
      "distillation_loss": 0.016880009323358536,
      "epoch": 7.68,
      "learning_rate": 4.886405581541986e-05,
      "loss": 0.0156,
      "step": 8086,
      "task_loss": 0.0044879671186208725
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999935405576537,
      "compression_loss": 0.0,
      "distillation_loss": 0.02322390116751194,
      "epoch": 7.68,
      "learning_rate": 4.886087826063252e-05,
      "loss": 0.0216,
      "step": 8087,
      "task_loss": 0.006913455203175545
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999935977205271,
      "compression_loss": 0.0,
      "distillation_loss": 0.055455997586250305,
      "epoch": 7.68,
      "learning_rate": 4.8857696371401735e-05,
      "loss": 0.0505,
      "step": 8088,
      "task_loss": 0.00570225715637207
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999936545451598,
      "compression_loss": 0.0,
      "distillation_loss": 0.28629836440086365,
      "epoch": 7.68,
      "learning_rate": 4.88545101483055e-05,
      "loss": 0.2723,
      "step": 8089,
      "task_loss": 0.14600038528442383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999937110325553,
      "compression_loss": 0.0,
      "distillation_loss": 0.04520022124052048,
      "epoch": 7.68,
      "learning_rate": 4.885131959192262e-05,
      "loss": 0.0451,
      "step": 8090,
      "task_loss": 0.0442010760307312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999937671837173,
      "compression_loss": 0.0,
      "distillation_loss": 0.0367523655295372,
      "epoch": 7.68,
      "learning_rate": 4.884812470283265e-05,
      "loss": 0.0401,
      "step": 8091,
      "task_loss": 0.07056266069412231
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999938229996495,
      "compression_loss": 0.0,
      "distillation_loss": 0.02565024048089981,
      "epoch": 7.68,
      "learning_rate": 4.884492548161599e-05,
      "loss": 0.0237,
      "step": 8092,
      "task_loss": 0.005736216902732849
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999938784813555,
      "compression_loss": 0.0,
      "distillation_loss": 0.14084674417972565,
      "epoch": 7.69,
      "learning_rate": 4.8841721928853776e-05,
      "loss": 0.142,
      "step": 8093,
      "task_loss": 0.15224382281303406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999939336298391,
      "compression_loss": 0.0,
      "distillation_loss": 0.04901870712637901,
      "epoch": 7.69,
      "learning_rate": 4.8838514045127945e-05,
      "loss": 0.0518,
      "step": 8094,
      "task_loss": 0.07676392048597336
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999939884461038,
      "compression_loss": 0.0,
      "distillation_loss": 0.07762657105922699,
      "epoch": 7.69,
      "learning_rate": 4.883530183102123e-05,
      "loss": 0.0828,
      "step": 8095,
      "task_loss": 0.1296282708644867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999940429311535,
      "compression_loss": 0.0,
      "distillation_loss": 0.060121599584817886,
      "epoch": 7.69,
      "learning_rate": 4.883208528711715e-05,
      "loss": 0.0675,
      "step": 8096,
      "task_loss": 0.13351008296012878
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999940970859918,
      "compression_loss": 0.0,
      "distillation_loss": 0.07687482237815857,
      "epoch": 7.69,
      "learning_rate": 4.8828864413999995e-05,
      "loss": 0.0771,
      "step": 8097,
      "task_loss": 0.07926255464553833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999941509116224,
      "compression_loss": 0.0,
      "distillation_loss": 0.06490933150053024,
      "epoch": 7.69,
      "learning_rate": 4.8825639212254865e-05,
      "loss": 0.0652,
      "step": 8098,
      "task_loss": 0.0676705464720726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999942044090489,
      "compression_loss": 0.0,
      "distillation_loss": 0.03710645064711571,
      "epoch": 7.69,
      "learning_rate": 4.882240968246762e-05,
      "loss": 0.0375,
      "step": 8099,
      "task_loss": 0.04076388105750084
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999942575792751,
      "compression_loss": 0.0,
      "distillation_loss": 0.02674899250268936,
      "epoch": 7.69,
      "learning_rate": 4.8819175825224925e-05,
      "loss": 0.0249,
      "step": 8100,
      "task_loss": 0.0087125264108181
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999943104233046,
      "compression_loss": 0.0,
      "distillation_loss": 0.06944914162158966,
      "epoch": 7.69,
      "learning_rate": 4.881593764111424e-05,
      "loss": 0.0701,
      "step": 8101,
      "task_loss": 0.07624385505914688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799994362942141,
      "compression_loss": 0.0,
      "distillation_loss": 0.015938565135002136,
      "epoch": 7.69,
      "learning_rate": 4.8812695130723775e-05,
      "loss": 0.0148,
      "step": 8102,
      "task_loss": 0.004970138892531395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999944151367882,
      "compression_loss": 0.0,
      "distillation_loss": 0.07653353363275528,
      "epoch": 7.7,
      "learning_rate": 4.880944829464256e-05,
      "loss": 0.0844,
      "step": 8103,
      "task_loss": 0.155586376786232
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999944670082497,
      "compression_loss": 0.0,
      "distillation_loss": 0.05173022672533989,
      "epoch": 7.7,
      "learning_rate": 4.880619713346039e-05,
      "loss": 0.0512,
      "step": 8104,
      "task_loss": 0.045994266867637634
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999945185575293,
      "compression_loss": 0.0,
      "distillation_loss": 0.04804886877536774,
      "epoch": 7.7,
      "learning_rate": 4.8802941647767856e-05,
      "loss": 0.0604,
      "step": 8105,
      "task_loss": 0.1717759370803833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999945697856307,
      "compression_loss": 0.0,
      "distillation_loss": 0.08852384239435196,
      "epoch": 7.7,
      "learning_rate": 4.879968183815634e-05,
      "loss": 0.0908,
      "step": 8106,
      "task_loss": 0.11161627620458603
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999946206935573,
      "compression_loss": 0.0,
      "distillation_loss": 0.11721711605787277,
      "epoch": 7.7,
      "learning_rate": 4.8796417705217994e-05,
      "loss": 0.1135,
      "step": 8107,
      "task_loss": 0.07980488240718842
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999946712823132,
      "compression_loss": 0.0,
      "distillation_loss": 0.14771784842014313,
      "epoch": 7.7,
      "learning_rate": 4.879314924954577e-05,
      "loss": 0.1429,
      "step": 8108,
      "task_loss": 0.09997382760047913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999947215529017,
      "compression_loss": 0.0,
      "distillation_loss": 0.07733377814292908,
      "epoch": 7.7,
      "learning_rate": 4.87898764717334e-05,
      "loss": 0.0812,
      "step": 8109,
      "task_loss": 0.1161360889673233
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999947715063268,
      "compression_loss": 0.0,
      "distillation_loss": 0.06381769478321075,
      "epoch": 7.7,
      "learning_rate": 4.8786599372375384e-05,
      "loss": 0.0612,
      "step": 8110,
      "task_loss": 0.03766363486647606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799994821143592,
      "compression_loss": 0.0,
      "distillation_loss": 0.16585132479667664,
      "epoch": 7.7,
      "learning_rate": 4.878331795206705e-05,
      "loss": 0.1596,
      "step": 8111,
      "task_loss": 0.10337799787521362
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799994870465701,
      "compression_loss": 0.0,
      "distillation_loss": 0.07819118350744247,
      "epoch": 7.7,
      "learning_rate": 4.878003221140446e-05,
      "loss": 0.0886,
      "step": 8112,
      "task_loss": 0.18247146904468536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999949194736575,
      "compression_loss": 0.0,
      "distillation_loss": 0.014311755076050758,
      "epoch": 7.7,
      "learning_rate": 4.877674215098449e-05,
      "loss": 0.0136,
      "step": 8113,
      "task_loss": 0.007542043924331665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999949681684653,
      "compression_loss": 0.0,
      "distillation_loss": 0.02407708764076233,
      "epoch": 7.71,
      "learning_rate": 4.87734477714048e-05,
      "loss": 0.0321,
      "step": 8114,
      "task_loss": 0.10392209887504578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999950165511278,
      "compression_loss": 0.0,
      "distillation_loss": 0.0483178049325943,
      "epoch": 7.71,
      "learning_rate": 4.8770149073263833e-05,
      "loss": 0.0544,
      "step": 8115,
      "task_loss": 0.10924448817968369
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999950646226489,
      "compression_loss": 0.0,
      "distillation_loss": 0.06605200469493866,
      "epoch": 7.71,
      "learning_rate": 4.87668460571608e-05,
      "loss": 0.0702,
      "step": 8116,
      "task_loss": 0.10719858109951019
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999951123840323,
      "compression_loss": 0.0,
      "distillation_loss": 0.12172948569059372,
      "epoch": 7.71,
      "learning_rate": 4.8763538723695726e-05,
      "loss": 0.132,
      "step": 8117,
      "task_loss": 0.22487123310565948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999951598362816,
      "compression_loss": 0.0,
      "distillation_loss": 0.037446945905685425,
      "epoch": 7.71,
      "learning_rate": 4.87602270734694e-05,
      "loss": 0.0366,
      "step": 8118,
      "task_loss": 0.028905699029564857
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999952069804004,
      "compression_loss": 0.0,
      "distillation_loss": 0.02980622835457325,
      "epoch": 7.71,
      "learning_rate": 4.8756911107083387e-05,
      "loss": 0.0321,
      "step": 8119,
      "task_loss": 0.05290570110082626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999952538173926,
      "compression_loss": 0.0,
      "distillation_loss": 0.02518196403980255,
      "epoch": 7.71,
      "learning_rate": 4.875359082514006e-05,
      "loss": 0.0235,
      "step": 8120,
      "task_loss": 0.008489791303873062
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999953003482617,
      "compression_loss": 0.0,
      "distillation_loss": 0.019510727375745773,
      "epoch": 7.71,
      "learning_rate": 4.8750266228242555e-05,
      "loss": 0.0322,
      "step": 8121,
      "task_loss": 0.14628659188747406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999953465740115,
      "compression_loss": 0.0,
      "distillation_loss": 0.0406201109290123,
      "epoch": 7.71,
      "learning_rate": 4.874693731699481e-05,
      "loss": 0.0407,
      "step": 8122,
      "task_loss": 0.04184962809085846
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999953924956456,
      "compression_loss": 0.0,
      "distillation_loss": 0.03238911181688309,
      "epoch": 7.71,
      "learning_rate": 4.8743604092001544e-05,
      "loss": 0.0395,
      "step": 8123,
      "task_loss": 0.10380247235298157
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999954381141676,
      "compression_loss": 0.0,
      "distillation_loss": 0.08192041516304016,
      "epoch": 7.72,
      "learning_rate": 4.8740266553868236e-05,
      "loss": 0.0867,
      "step": 8124,
      "task_loss": 0.12997817993164062
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999954834305814,
      "compression_loss": 0.0,
      "distillation_loss": 0.040653470903635025,
      "epoch": 7.72,
      "learning_rate": 4.873692470320117e-05,
      "loss": 0.0513,
      "step": 8125,
      "task_loss": 0.14687559008598328
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999955284458905,
      "compression_loss": 0.0,
      "distillation_loss": 0.07809877395629883,
      "epoch": 7.72,
      "learning_rate": 4.8733578540607425e-05,
      "loss": 0.0812,
      "step": 8126,
      "task_loss": 0.10933775454759598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999955731610988,
      "compression_loss": 0.0,
      "distillation_loss": 0.0384988859295845,
      "epoch": 7.72,
      "learning_rate": 4.8730228066694825e-05,
      "loss": 0.0491,
      "step": 8127,
      "task_loss": 0.1440470814704895
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999956175772097,
      "compression_loss": 0.0,
      "distillation_loss": 0.015554778277873993,
      "epoch": 7.72,
      "learning_rate": 4.872687328207202e-05,
      "loss": 0.0146,
      "step": 8128,
      "task_loss": 0.006275909021496773
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999956616952271,
      "compression_loss": 0.0,
      "distillation_loss": 0.013756733387708664,
      "epoch": 7.72,
      "learning_rate": 4.872351418734841e-05,
      "loss": 0.0128,
      "step": 8129,
      "task_loss": 0.004298551008105278
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999957055161545,
      "compression_loss": 0.0,
      "distillation_loss": 0.037815727293491364,
      "epoch": 7.72,
      "learning_rate": 4.8720150783134196e-05,
      "loss": 0.0426,
      "step": 8130,
      "task_loss": 0.08522268384695053
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999957490409958,
      "compression_loss": 0.0,
      "distillation_loss": 0.0777861624956131,
      "epoch": 7.72,
      "learning_rate": 4.871678307004035e-05,
      "loss": 0.0729,
      "step": 8131,
      "task_loss": 0.0290694423019886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999957922707545,
      "compression_loss": 0.0,
      "distillation_loss": 0.021506596356630325,
      "epoch": 7.72,
      "learning_rate": 4.8713411048678635e-05,
      "loss": 0.0198,
      "step": 8132,
      "task_loss": 0.004600740969181061
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999958352064344,
      "compression_loss": 0.0,
      "distillation_loss": 0.13485758006572723,
      "epoch": 7.72,
      "learning_rate": 4.8710034719661614e-05,
      "loss": 0.1337,
      "step": 8133,
      "task_loss": 0.1231408417224884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799995877849039,
      "compression_loss": 0.0,
      "distillation_loss": 0.014092806726694107,
      "epoch": 7.72,
      "learning_rate": 4.870665408360258e-05,
      "loss": 0.0216,
      "step": 8134,
      "task_loss": 0.08945970237255096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999959201995722,
      "compression_loss": 0.0,
      "distillation_loss": 0.04725116491317749,
      "epoch": 7.73,
      "learning_rate": 4.870326914111567e-05,
      "loss": 0.044,
      "step": 8135,
      "task_loss": 0.014543771743774414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999959622590376,
      "compression_loss": 0.0,
      "distillation_loss": 0.09922022372484207,
      "epoch": 7.73,
      "learning_rate": 4.8699879892815756e-05,
      "loss": 0.09,
      "step": 8136,
      "task_loss": 0.007207704707980156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999960040284388,
      "compression_loss": 0.0,
      "distillation_loss": 0.062455762177705765,
      "epoch": 7.73,
      "learning_rate": 4.8696486339318524e-05,
      "loss": 0.0625,
      "step": 8137,
      "task_loss": 0.06264299154281616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999960455087797,
      "compression_loss": 0.0,
      "distillation_loss": 0.08884945511817932,
      "epoch": 7.73,
      "learning_rate": 4.8693088481240424e-05,
      "loss": 0.0883,
      "step": 8138,
      "task_loss": 0.0832391083240509
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999960867010637,
      "compression_loss": 0.0,
      "distillation_loss": 0.03622208908200264,
      "epoch": 7.73,
      "learning_rate": 4.86896863191987e-05,
      "loss": 0.0334,
      "step": 8139,
      "task_loss": 0.007611839100718498
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999961276062947,
      "compression_loss": 0.0,
      "distillation_loss": 0.05895059183239937,
      "epoch": 7.73,
      "learning_rate": 4.8686279853811356e-05,
      "loss": 0.0593,
      "step": 8140,
      "task_loss": 0.06226490065455437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999961682254764,
      "compression_loss": 0.0,
      "distillation_loss": 0.03609544038772583,
      "epoch": 7.73,
      "learning_rate": 4.8682869085697206e-05,
      "loss": 0.0412,
      "step": 8141,
      "task_loss": 0.0870952308177948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999962085596122,
      "compression_loss": 0.0,
      "distillation_loss": 0.09176631271839142,
      "epoch": 7.73,
      "learning_rate": 4.8679454015475835e-05,
      "loss": 0.0958,
      "step": 8142,
      "task_loss": 0.13180381059646606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999962486097061,
      "compression_loss": 0.0,
      "distillation_loss": 0.15721860527992249,
      "epoch": 7.73,
      "learning_rate": 4.867603464376759e-05,
      "loss": 0.1509,
      "step": 8143,
      "task_loss": 0.09391912072896957
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999962883767615,
      "compression_loss": 0.0,
      "distillation_loss": 0.020519878715276718,
      "epoch": 7.73,
      "learning_rate": 4.867261097119363e-05,
      "loss": 0.0241,
      "step": 8144,
      "task_loss": 0.0560954324901104
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999963278617823,
      "compression_loss": 0.0,
      "distillation_loss": 0.10751358419656754,
      "epoch": 7.74,
      "learning_rate": 4.8669182998375884e-05,
      "loss": 0.1109,
      "step": 8145,
      "task_loss": 0.14114311337471008
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999963670657722,
      "compression_loss": 0.0,
      "distillation_loss": 0.06851037591695786,
      "epoch": 7.74,
      "learning_rate": 4.8665750725937045e-05,
      "loss": 0.0684,
      "step": 8146,
      "task_loss": 0.06784656643867493
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999964059897348,
      "compression_loss": 0.0,
      "distillation_loss": 0.0376189686357975,
      "epoch": 7.74,
      "learning_rate": 4.866231415450062e-05,
      "loss": 0.0377,
      "step": 8147,
      "task_loss": 0.03808595612645149
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999964446346736,
      "compression_loss": 0.0,
      "distillation_loss": 0.03249119967222214,
      "epoch": 7.74,
      "learning_rate": 4.8658873284690866e-05,
      "loss": 0.0449,
      "step": 8148,
      "task_loss": 0.15644009411334991
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999964830015925,
      "compression_loss": 0.0,
      "distillation_loss": 0.04549805074930191,
      "epoch": 7.74,
      "learning_rate": 4.865542811713284e-05,
      "loss": 0.0521,
      "step": 8149,
      "task_loss": 0.11196024715900421
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999965210914953,
      "compression_loss": 0.0,
      "distillation_loss": 0.07704181969165802,
      "epoch": 7.74,
      "learning_rate": 4.865197865245237e-05,
      "loss": 0.0857,
      "step": 8150,
      "task_loss": 0.16380248963832855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999965589053855,
      "compression_loss": 0.0,
      "distillation_loss": 0.06383125483989716,
      "epoch": 7.74,
      "learning_rate": 4.8648524891276066e-05,
      "loss": 0.0624,
      "step": 8151,
      "task_loss": 0.049697332084178925
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999965964442667,
      "compression_loss": 0.0,
      "distillation_loss": 0.10283863544464111,
      "epoch": 7.74,
      "learning_rate": 4.8645066834231325e-05,
      "loss": 0.1101,
      "step": 8152,
      "task_loss": 0.17590855062007904
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999966337091426,
      "compression_loss": 0.0,
      "distillation_loss": 0.09706176817417145,
      "epoch": 7.74,
      "learning_rate": 4.8641604481946314e-05,
      "loss": 0.1095,
      "step": 8153,
      "task_loss": 0.22108034789562225
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999966707010172,
      "compression_loss": 0.0,
      "distillation_loss": 0.08932052552700043,
      "epoch": 7.74,
      "learning_rate": 4.863813783504999e-05,
      "loss": 0.0871,
      "step": 8154,
      "task_loss": 0.0668034479022026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999967074208939,
      "compression_loss": 0.0,
      "distillation_loss": 0.022766336798667908,
      "epoch": 7.74,
      "learning_rate": 4.863466689417209e-05,
      "loss": 0.0263,
      "step": 8155,
      "task_loss": 0.058080703020095825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999967438697764,
      "compression_loss": 0.0,
      "distillation_loss": 0.04971291124820709,
      "epoch": 7.75,
      "learning_rate": 4.863119165994312e-05,
      "loss": 0.0537,
      "step": 8156,
      "task_loss": 0.08949106186628342
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999967800486684,
      "compression_loss": 0.0,
      "distillation_loss": 0.04288540780544281,
      "epoch": 7.75,
      "learning_rate": 4.862771213299438e-05,
      "loss": 0.039,
      "step": 8157,
      "task_loss": 0.004233557730913162
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999968159585735,
      "compression_loss": 0.0,
      "distillation_loss": 0.01929197832942009,
      "epoch": 7.75,
      "learning_rate": 4.8624228313957937e-05,
      "loss": 0.0265,
      "step": 8158,
      "task_loss": 0.09141400456428528
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999968516004956,
      "compression_loss": 0.0,
      "distillation_loss": 0.04137708246707916,
      "epoch": 7.75,
      "learning_rate": 4.862074020346664e-05,
      "loss": 0.0378,
      "step": 8159,
      "task_loss": 0.0055605582892894745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999968869754382,
      "compression_loss": 0.0,
      "distillation_loss": 0.01853022351861,
      "epoch": 7.75,
      "learning_rate": 4.8617247802154134e-05,
      "loss": 0.0252,
      "step": 8160,
      "task_loss": 0.085117407143116
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999969220844052,
      "compression_loss": 0.0,
      "distillation_loss": 0.07992290705442429,
      "epoch": 7.75,
      "learning_rate": 4.861375111065482e-05,
      "loss": 0.0841,
      "step": 8161,
      "task_loss": 0.12121039628982544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999969569284,
      "compression_loss": 0.0,
      "distillation_loss": 0.02700880542397499,
      "epoch": 7.75,
      "learning_rate": 4.861025012960389e-05,
      "loss": 0.0271,
      "step": 8162,
      "task_loss": 0.027840938419103622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999969915084264,
      "compression_loss": 0.0,
      "distillation_loss": 0.04433220624923706,
      "epoch": 7.75,
      "learning_rate": 4.8606744859637316e-05,
      "loss": 0.0487,
      "step": 8163,
      "task_loss": 0.08848340809345245
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999970258254882,
      "compression_loss": 0.0,
      "distillation_loss": 0.022559959441423416,
      "epoch": 7.75,
      "learning_rate": 4.8603235301391844e-05,
      "loss": 0.0306,
      "step": 8164,
      "task_loss": 0.10257212072610855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999970598805889,
      "compression_loss": 0.0,
      "distillation_loss": 0.020599160343408585,
      "epoch": 7.75,
      "learning_rate": 4.859972145550501e-05,
      "loss": 0.0287,
      "step": 8165,
      "task_loss": 0.10190172493457794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999970936747323,
      "compression_loss": 0.0,
      "distillation_loss": 0.044832177460193634,
      "epoch": 7.75,
      "learning_rate": 4.859620332261512e-05,
      "loss": 0.0469,
      "step": 8166,
      "task_loss": 0.0659225732088089
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999971272089219,
      "compression_loss": 0.0,
      "distillation_loss": 0.0355226993560791,
      "epoch": 7.76,
      "learning_rate": 4.8592680903361247e-05,
      "loss": 0.0403,
      "step": 8167,
      "task_loss": 0.08283548802137375
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999971604841616,
      "compression_loss": 0.0,
      "distillation_loss": 0.06762556731700897,
      "epoch": 7.76,
      "learning_rate": 4.858915419838327e-05,
      "loss": 0.0614,
      "step": 8168,
      "task_loss": 0.005646536126732826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799997193501455,
      "compression_loss": 0.0,
      "distillation_loss": 0.02109435945749283,
      "epoch": 7.76,
      "learning_rate": 4.8585623208321825e-05,
      "loss": 0.0389,
      "step": 8169,
      "task_loss": 0.19891361892223358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999972262618058,
      "compression_loss": 0.0,
      "distillation_loss": 0.029250595718622208,
      "epoch": 7.76,
      "learning_rate": 4.858208793381833e-05,
      "loss": 0.0603,
      "step": 8170,
      "task_loss": 0.3400185704231262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999972587662176,
      "compression_loss": 0.0,
      "distillation_loss": 0.10609747469425201,
      "epoch": 7.76,
      "learning_rate": 4.8578548375514995e-05,
      "loss": 0.0997,
      "step": 8171,
      "task_loss": 0.042343318462371826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999972910156943,
      "compression_loss": 0.0,
      "distillation_loss": 0.03819608315825462,
      "epoch": 7.76,
      "learning_rate": 4.8575004534054794e-05,
      "loss": 0.036,
      "step": 8172,
      "task_loss": 0.016501694917678833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999973230112394,
      "compression_loss": 0.0,
      "distillation_loss": 0.020714174956083298,
      "epoch": 7.76,
      "learning_rate": 4.8571456410081474e-05,
      "loss": 0.0334,
      "step": 8173,
      "task_loss": 0.14768551290035248
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999973547538565,
      "compression_loss": 0.0,
      "distillation_loss": 0.04434049874544144,
      "epoch": 7.76,
      "learning_rate": 4.856790400423958e-05,
      "loss": 0.0539,
      "step": 8174,
      "task_loss": 0.13957616686820984
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999973862445494,
      "compression_loss": 0.0,
      "distillation_loss": 0.024984436109662056,
      "epoch": 7.76,
      "learning_rate": 4.856434731717442e-05,
      "loss": 0.0395,
      "step": 8175,
      "task_loss": 0.1703425943851471
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999974174843218,
      "compression_loss": 0.0,
      "distillation_loss": 0.013447124511003494,
      "epoch": 7.76,
      "learning_rate": 4.8560786349532075e-05,
      "loss": 0.0169,
      "step": 8176,
      "task_loss": 0.04839882627129555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999974484741774,
      "compression_loss": 0.0,
      "distillation_loss": 0.04491272196173668,
      "epoch": 7.77,
      "learning_rate": 4.855722110195943e-05,
      "loss": 0.0685,
      "step": 8177,
      "task_loss": 0.28108105063438416
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999974792151199,
      "compression_loss": 0.0,
      "distillation_loss": 0.022170700132846832,
      "epoch": 7.77,
      "learning_rate": 4.8553651575104114e-05,
      "loss": 0.033,
      "step": 8178,
      "task_loss": 0.130245178937912
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999975097081529,
      "compression_loss": 0.0,
      "distillation_loss": 0.0931999534368515,
      "epoch": 7.77,
      "learning_rate": 4.8550077769614554e-05,
      "loss": 0.0971,
      "step": 8179,
      "task_loss": 0.1317388266324997
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79999753995428,
      "compression_loss": 0.0,
      "distillation_loss": 0.019387193024158478,
      "epoch": 7.77,
      "learning_rate": 4.8546499686139944e-05,
      "loss": 0.0178,
      "step": 8180,
      "task_loss": 0.0039298199117183685
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799997569954505,
      "compression_loss": 0.0,
      "distillation_loss": 0.014103731140494347,
      "epoch": 7.77,
      "learning_rate": 4.854291732533027e-05,
      "loss": 0.0132,
      "step": 8181,
      "task_loss": 0.005378361791372299
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999975997098316,
      "compression_loss": 0.0,
      "distillation_loss": 0.019791752099990845,
      "epoch": 7.77,
      "learning_rate": 4.853933068783628e-05,
      "loss": 0.0279,
      "step": 8182,
      "task_loss": 0.10089607536792755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999976292212635,
      "compression_loss": 0.0,
      "distillation_loss": 0.08148065209388733,
      "epoch": 7.77,
      "learning_rate": 4.853573977430951e-05,
      "loss": 0.0862,
      "step": 8183,
      "task_loss": 0.128190815448761
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999976584898043,
      "compression_loss": 0.0,
      "distillation_loss": 0.045781854540109634,
      "epoch": 7.77,
      "learning_rate": 4.8532144585402254e-05,
      "loss": 0.0464,
      "step": 8184,
      "task_loss": 0.051745254546403885
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999976875164577,
      "compression_loss": 0.0,
      "distillation_loss": 0.0476515106856823,
      "epoch": 7.77,
      "learning_rate": 4.85285451217676e-05,
      "loss": 0.0505,
      "step": 8185,
      "task_loss": 0.07574954628944397
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999977163022274,
      "compression_loss": 0.0,
      "distillation_loss": 0.15082959830760956,
      "epoch": 7.77,
      "learning_rate": 4.8524941384059415e-05,
      "loss": 0.1472,
      "step": 8186,
      "task_loss": 0.11474103480577469
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999977448481171,
      "compression_loss": 0.0,
      "distillation_loss": 0.016797035932540894,
      "epoch": 7.77,
      "learning_rate": 4.8521333372932326e-05,
      "loss": 0.0269,
      "step": 8187,
      "task_loss": 0.11737027764320374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999977731551304,
      "compression_loss": 0.0,
      "distillation_loss": 0.035971708595752716,
      "epoch": 7.78,
      "learning_rate": 4.851772108904175e-05,
      "loss": 0.0345,
      "step": 8188,
      "task_loss": 0.021439943462610245
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999978012242711,
      "compression_loss": 0.0,
      "distillation_loss": 0.024865467101335526,
      "epoch": 7.78,
      "learning_rate": 4.851410453304388e-05,
      "loss": 0.0321,
      "step": 8189,
      "task_loss": 0.09683802723884583
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999978290565428,
      "compression_loss": 0.0,
      "distillation_loss": 0.02018059231340885,
      "epoch": 7.78,
      "learning_rate": 4.851048370559567e-05,
      "loss": 0.0277,
      "step": 8190,
      "task_loss": 0.09545273333787918
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999978566529493,
      "compression_loss": 0.0,
      "distillation_loss": 0.032620202749967575,
      "epoch": 7.78,
      "learning_rate": 4.850685860735487e-05,
      "loss": 0.0374,
      "step": 8191,
      "task_loss": 0.08022205531597137
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999978840144941,
      "compression_loss": 0.0,
      "distillation_loss": 0.03163286671042442,
      "epoch": 7.78,
      "learning_rate": 4.850322923898e-05,
      "loss": 0.0294,
      "step": 8192,
      "task_loss": 0.00926467590034008
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799997911142181,
      "compression_loss": 0.0,
      "distillation_loss": 0.06667460501194,
      "epoch": 7.78,
      "learning_rate": 4.8499595601130337e-05,
      "loss": 0.0806,
      "step": 8193,
      "task_loss": 0.20638611912727356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999979380370137,
      "compression_loss": 0.0,
      "distillation_loss": 0.1302226334810257,
      "epoch": 7.78,
      "learning_rate": 4.849595769446596e-05,
      "loss": 0.1247,
      "step": 8194,
      "task_loss": 0.07523670792579651
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999979646999957,
      "compression_loss": 0.0,
      "distillation_loss": 0.12275674939155579,
      "epoch": 7.78,
      "learning_rate": 4.849231551964771e-05,
      "loss": 0.1306,
      "step": 8195,
      "task_loss": 0.20129883289337158
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999979911321309,
      "compression_loss": 0.0,
      "distillation_loss": 0.03874664753675461,
      "epoch": 7.78,
      "learning_rate": 4.848866907733721e-05,
      "loss": 0.0359,
      "step": 8196,
      "task_loss": 0.009907728061079979
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999980173344229,
      "compression_loss": 0.0,
      "distillation_loss": 0.07732771337032318,
      "epoch": 7.78,
      "learning_rate": 4.848501836819684e-05,
      "loss": 0.0893,
      "step": 8197,
      "task_loss": 0.19742365181446075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999980433078754,
      "compression_loss": 0.0,
      "distillation_loss": 0.024056117981672287,
      "epoch": 7.79,
      "learning_rate": 4.848136339288979e-05,
      "loss": 0.0229,
      "step": 8198,
      "task_loss": 0.012888273224234581
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799998069053492,
      "compression_loss": 0.0,
      "distillation_loss": 0.1355004906654358,
      "epoch": 7.79,
      "learning_rate": 4.8477704152079984e-05,
      "loss": 0.1258,
      "step": 8199,
      "task_loss": 0.03880665823817253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999980945722766,
      "compression_loss": 0.0,
      "distillation_loss": 0.09072038531303406,
      "epoch": 7.79,
      "learning_rate": 4.8474040646432153e-05,
      "loss": 0.0949,
      "step": 8200,
      "task_loss": 0.13292811810970306
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999981198652326,
      "compression_loss": 0.0,
      "distillation_loss": 0.04228777810931206,
      "epoch": 7.79,
      "learning_rate": 4.8470372876611784e-05,
      "loss": 0.0399,
      "step": 8201,
      "task_loss": 0.0188329815864563
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999981449333639,
      "compression_loss": 0.0,
      "distillation_loss": 0.05842405557632446,
      "epoch": 7.79,
      "learning_rate": 4.846670084328515e-05,
      "loss": 0.0558,
      "step": 8202,
      "task_loss": 0.03242020681500435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999981697776739,
      "compression_loss": 0.0,
      "distillation_loss": 0.10038132965564728,
      "epoch": 7.79,
      "learning_rate": 4.846302454711929e-05,
      "loss": 0.0992,
      "step": 8203,
      "task_loss": 0.08900587260723114
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999981943991666,
      "compression_loss": 0.0,
      "distillation_loss": 0.06294578313827515,
      "epoch": 7.79,
      "learning_rate": 4.845934398878202e-05,
      "loss": 0.0766,
      "step": 8204,
      "task_loss": 0.19954392313957214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999982187988456,
      "compression_loss": 0.0,
      "distillation_loss": 0.015763722360134125,
      "epoch": 7.79,
      "learning_rate": 4.845565916894193e-05,
      "loss": 0.0244,
      "step": 8205,
      "task_loss": 0.10191762447357178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999982429777145,
      "compression_loss": 0.0,
      "distillation_loss": 0.04905875027179718,
      "epoch": 7.79,
      "learning_rate": 4.8451970088268396e-05,
      "loss": 0.0462,
      "step": 8206,
      "task_loss": 0.0201642494648695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799998266936777,
      "compression_loss": 0.0,
      "distillation_loss": 0.03759082779288292,
      "epoch": 7.79,
      "learning_rate": 4.8448276747431545e-05,
      "loss": 0.0349,
      "step": 8207,
      "task_loss": 0.010456247255206108
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999982906770369,
      "compression_loss": 0.0,
      "distillation_loss": 0.01988394558429718,
      "epoch": 7.79,
      "learning_rate": 4.84445791471023e-05,
      "loss": 0.0286,
      "step": 8208,
      "task_loss": 0.10685549676418304
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999983141994976,
      "compression_loss": 0.0,
      "distillation_loss": 0.1849537491798401,
      "epoch": 7.8,
      "learning_rate": 4.8440877287952336e-05,
      "loss": 0.1814,
      "step": 8209,
      "task_loss": 0.14974652230739594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999983375051631,
      "compression_loss": 0.0,
      "distillation_loss": 0.03731781989336014,
      "epoch": 7.8,
      "learning_rate": 4.8437171170654125e-05,
      "loss": 0.0406,
      "step": 8210,
      "task_loss": 0.07024012506008148
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799998360595037,
      "compression_loss": 0.0,
      "distillation_loss": 0.027419112622737885,
      "epoch": 7.8,
      "learning_rate": 4.843346079588089e-05,
      "loss": 0.0254,
      "step": 8211,
      "task_loss": 0.00690159946680069
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999983834701229,
      "compression_loss": 0.0,
      "distillation_loss": 0.03823021054267883,
      "epoch": 7.8,
      "learning_rate": 4.842974616430665e-05,
      "loss": 0.0443,
      "step": 8212,
      "task_loss": 0.09891167283058167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999984061314245,
      "compression_loss": 0.0,
      "distillation_loss": 0.022461410611867905,
      "epoch": 7.8,
      "learning_rate": 4.842602727660618e-05,
      "loss": 0.0244,
      "step": 8213,
      "task_loss": 0.04227849468588829
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999984285799454,
      "compression_loss": 0.0,
      "distillation_loss": 0.06684540212154388,
      "epoch": 7.8,
      "learning_rate": 4.842230413345503e-05,
      "loss": 0.0692,
      "step": 8214,
      "task_loss": 0.09044404327869415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999984508166894,
      "compression_loss": 0.0,
      "distillation_loss": 0.041791971772909164,
      "epoch": 7.8,
      "learning_rate": 4.8418576735529535e-05,
      "loss": 0.0441,
      "step": 8215,
      "task_loss": 0.06479668617248535
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999984728426602,
      "compression_loss": 0.0,
      "distillation_loss": 0.03619157522916794,
      "epoch": 7.8,
      "learning_rate": 4.841484508350679e-05,
      "loss": 0.0353,
      "step": 8216,
      "task_loss": 0.02702299691736698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999984946588615,
      "compression_loss": 0.0,
      "distillation_loss": 0.0437760166823864,
      "epoch": 7.8,
      "learning_rate": 4.841110917806467e-05,
      "loss": 0.0542,
      "step": 8217,
      "task_loss": 0.14790481328964233
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999985162662969,
      "compression_loss": 0.0,
      "distillation_loss": 0.03585965931415558,
      "epoch": 7.8,
      "learning_rate": 4.840736901988182e-05,
      "loss": 0.0396,
      "step": 8218,
      "task_loss": 0.07364504784345627
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999985376659701,
      "compression_loss": 0.0,
      "distillation_loss": 0.12400738894939423,
      "epoch": 7.81,
      "learning_rate": 4.840362460963765e-05,
      "loss": 0.1329,
      "step": 8219,
      "task_loss": 0.21300512552261353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999985588588847,
      "compression_loss": 0.0,
      "distillation_loss": 0.04792033135890961,
      "epoch": 7.81,
      "learning_rate": 4.8399875948012355e-05,
      "loss": 0.0518,
      "step": 8220,
      "task_loss": 0.0868147611618042
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999985798460446,
      "compression_loss": 0.0,
      "distillation_loss": 0.04659024253487587,
      "epoch": 7.81,
      "learning_rate": 4.8396123035686906e-05,
      "loss": 0.0427,
      "step": 8221,
      "task_loss": 0.007754124701023102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999986006284533,
      "compression_loss": 0.0,
      "distillation_loss": 0.19927890598773956,
      "epoch": 7.81,
      "learning_rate": 4.839236587334303e-05,
      "loss": 0.2014,
      "step": 8222,
      "task_loss": 0.22048905491828918
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999986212071145,
      "compression_loss": 0.0,
      "distillation_loss": 0.08156438916921616,
      "epoch": 7.81,
      "learning_rate": 4.8388604461663236e-05,
      "loss": 0.0814,
      "step": 8223,
      "task_loss": 0.08026103675365448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799998641583032,
      "compression_loss": 0.0,
      "distillation_loss": 0.03177830949425697,
      "epoch": 7.81,
      "learning_rate": 4.838483880133079e-05,
      "loss": 0.0392,
      "step": 8224,
      "task_loss": 0.10596133768558502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999986617572094,
      "compression_loss": 0.0,
      "distillation_loss": 0.23754796385765076,
      "epoch": 7.81,
      "learning_rate": 4.8381068893029766e-05,
      "loss": 0.2324,
      "step": 8225,
      "task_loss": 0.18612830340862274
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999986817306503,
      "compression_loss": 0.0,
      "distillation_loss": 0.04634622856974602,
      "epoch": 7.81,
      "learning_rate": 4.837729473744497e-05,
      "loss": 0.0424,
      "step": 8226,
      "task_loss": 0.0071801114827394485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999987015043585,
      "compression_loss": 0.0,
      "distillation_loss": 0.03712467849254608,
      "epoch": 7.81,
      "learning_rate": 4.8373516335261994e-05,
      "loss": 0.034,
      "step": 8227,
      "task_loss": 0.0057260747998952866
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999987210793376,
      "compression_loss": 0.0,
      "distillation_loss": 0.038928285241127014,
      "epoch": 7.81,
      "learning_rate": 4.8369733687167204e-05,
      "loss": 0.0372,
      "step": 8228,
      "task_loss": 0.021699724718928337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999987404565914,
      "compression_loss": 0.0,
      "distillation_loss": 0.022505152970552444,
      "epoch": 7.81,
      "learning_rate": 4.836594679384775e-05,
      "loss": 0.0259,
      "step": 8229,
      "task_loss": 0.056766681373119354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999987596371234,
      "compression_loss": 0.0,
      "distillation_loss": 0.10139244794845581,
      "epoch": 7.82,
      "learning_rate": 4.836215565599152e-05,
      "loss": 0.096,
      "step": 8230,
      "task_loss": 0.0474693700671196
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999987786219375,
      "compression_loss": 0.0,
      "distillation_loss": 0.14026379585266113,
      "epoch": 7.82,
      "learning_rate": 4.835836027428722e-05,
      "loss": 0.1379,
      "step": 8231,
      "task_loss": 0.11672936379909515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999987974120373,
      "compression_loss": 0.0,
      "distillation_loss": 0.10862313210964203,
      "epoch": 7.82,
      "learning_rate": 4.8354560649424264e-05,
      "loss": 0.1071,
      "step": 8232,
      "task_loss": 0.09331917762756348
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999988160084264,
      "compression_loss": 0.0,
      "distillation_loss": 0.01986592821776867,
      "epoch": 7.82,
      "learning_rate": 4.8350756782092894e-05,
      "loss": 0.026,
      "step": 8233,
      "task_loss": 0.08163642883300781
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999988344121085,
      "compression_loss": 0.0,
      "distillation_loss": 0.05118180066347122,
      "epoch": 7.82,
      "learning_rate": 4.8346948672984096e-05,
      "loss": 0.0483,
      "step": 8234,
      "task_loss": 0.0220907311886549
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999988526240873,
      "compression_loss": 0.0,
      "distillation_loss": 0.14736244082450867,
      "epoch": 7.82,
      "learning_rate": 4.8343136322789626e-05,
      "loss": 0.1584,
      "step": 8235,
      "task_loss": 0.2577117383480072
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999988706453666,
      "compression_loss": 0.0,
      "distillation_loss": 0.1561247706413269,
      "epoch": 7.82,
      "learning_rate": 4.8339319732202024e-05,
      "loss": 0.1704,
      "step": 8236,
      "task_loss": 0.29920265078544617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79999888847695,
      "compression_loss": 0.0,
      "distillation_loss": 0.02194085717201233,
      "epoch": 7.82,
      "learning_rate": 4.83354989019146e-05,
      "loss": 0.0284,
      "step": 8237,
      "task_loss": 0.0863027274608612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799998906119841,
      "compression_loss": 0.0,
      "distillation_loss": 0.012062850408256054,
      "epoch": 7.82,
      "learning_rate": 4.83316738326214e-05,
      "loss": 0.0113,
      "step": 8238,
      "task_loss": 0.004339534789323807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999989235750437,
      "compression_loss": 0.0,
      "distillation_loss": 0.11506865918636322,
      "epoch": 7.82,
      "learning_rate": 4.832784452501729e-05,
      "loss": 0.1089,
      "step": 8239,
      "task_loss": 0.05316751450300217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999989408435614,
      "compression_loss": 0.0,
      "distillation_loss": 0.02399417757987976,
      "epoch": 7.83,
      "learning_rate": 4.8324010979797875e-05,
      "loss": 0.0228,
      "step": 8240,
      "task_loss": 0.012368382886052132
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999989579263979,
      "compression_loss": 0.0,
      "distillation_loss": 0.08361876010894775,
      "epoch": 7.83,
      "learning_rate": 4.8320173197659534e-05,
      "loss": 0.0955,
      "step": 8241,
      "task_loss": 0.20286312699317932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999989748245568,
      "compression_loss": 0.0,
      "distillation_loss": 0.03970951959490776,
      "epoch": 7.83,
      "learning_rate": 4.831633117929942e-05,
      "loss": 0.0364,
      "step": 8242,
      "task_loss": 0.006229208782315254
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799998991539042,
      "compression_loss": 0.0,
      "distillation_loss": 0.01506102830171585,
      "epoch": 7.83,
      "learning_rate": 4.831248492541545e-05,
      "loss": 0.0141,
      "step": 8243,
      "task_loss": 0.005801372230052948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999008070857,
      "compression_loss": 0.0,
      "distillation_loss": 0.040881797671318054,
      "epoch": 7.83,
      "learning_rate": 4.830863443670632e-05,
      "loss": 0.0392,
      "step": 8244,
      "task_loss": 0.023911267518997192
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999990244210056,
      "compression_loss": 0.0,
      "distillation_loss": 0.01167452521622181,
      "epoch": 7.83,
      "learning_rate": 4.8304779713871495e-05,
      "loss": 0.0109,
      "step": 8245,
      "task_loss": 0.0038937367498874664
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999990405904914,
      "compression_loss": 0.0,
      "distillation_loss": 0.15877902507781982,
      "epoch": 7.83,
      "learning_rate": 4.83009207576112e-05,
      "loss": 0.1789,
      "step": 8246,
      "task_loss": 0.3599551320075989
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999990565803181,
      "compression_loss": 0.0,
      "distillation_loss": 0.03877423331141472,
      "epoch": 7.83,
      "learning_rate": 4.829705756862642e-05,
      "loss": 0.0432,
      "step": 8247,
      "task_loss": 0.08290007710456848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999990723914894,
      "compression_loss": 0.0,
      "distillation_loss": 0.12737919390201569,
      "epoch": 7.83,
      "learning_rate": 4.829319014761894e-05,
      "loss": 0.1222,
      "step": 8248,
      "task_loss": 0.07570772618055344
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999088025009,
      "compression_loss": 0.0,
      "distillation_loss": 0.07020660489797592,
      "epoch": 7.83,
      "learning_rate": 4.828931849529129e-05,
      "loss": 0.0701,
      "step": 8249,
      "task_loss": 0.06908702105283737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999991034818805,
      "compression_loss": 0.0,
      "distillation_loss": 0.10183189064264297,
      "epoch": 7.83,
      "learning_rate": 4.8285442612346774e-05,
      "loss": 0.0971,
      "step": 8250,
      "task_loss": 0.05413543060421944
    },
    {
      "epoch": 7.83,
      "eval_accuracy": 0.8899082568807339,
      "eval_loss": 0.4171345829963684,
      "eval_runtime": 18.0817,
      "eval_samples_per_second": 48.225,
      "eval_steps_per_second": 6.028,
      "step": 8250
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999991187631076,
      "compression_loss": 0.0,
      "distillation_loss": 0.07504773885011673,
      "epoch": 7.84,
      "learning_rate": 4.828156249948946e-05,
      "loss": 0.0727,
      "step": 8251,
      "task_loss": 0.051514700055122375
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999991338696941,
      "compression_loss": 0.0,
      "distillation_loss": 0.022689666599035263,
      "epoch": 7.84,
      "learning_rate": 4.827767815742419e-05,
      "loss": 0.031,
      "step": 8252,
      "task_loss": 0.10627881437540054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999991488026436,
      "compression_loss": 0.0,
      "distillation_loss": 0.045913174748420715,
      "epoch": 7.84,
      "learning_rate": 4.8273789586856574e-05,
      "loss": 0.0693,
      "step": 8253,
      "task_loss": 0.27937787771224976
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999991635629597,
      "compression_loss": 0.0,
      "distillation_loss": 0.08047342300415039,
      "epoch": 7.84,
      "learning_rate": 4.8269896788493e-05,
      "loss": 0.0857,
      "step": 8254,
      "task_loss": 0.1325674057006836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999991781516462,
      "compression_loss": 0.0,
      "distillation_loss": 0.02534073404967785,
      "epoch": 7.84,
      "learning_rate": 4.8265999763040603e-05,
      "loss": 0.0333,
      "step": 8255,
      "task_loss": 0.10540292412042618
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999991925697068,
      "compression_loss": 0.0,
      "distillation_loss": 0.08169533312320709,
      "epoch": 7.84,
      "learning_rate": 4.8262098511207295e-05,
      "loss": 0.0787,
      "step": 8256,
      "task_loss": 0.0512889064848423
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999992068181451,
      "compression_loss": 0.0,
      "distillation_loss": 0.14142774045467377,
      "epoch": 7.84,
      "learning_rate": 4.825819303370177e-05,
      "loss": 0.1345,
      "step": 8257,
      "task_loss": 0.07237007468938828
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999992208979647,
      "compression_loss": 0.0,
      "distillation_loss": 0.026454931125044823,
      "epoch": 7.84,
      "learning_rate": 4.8254283331233464e-05,
      "loss": 0.0294,
      "step": 8258,
      "task_loss": 0.05604104697704315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999992348101694,
      "compression_loss": 0.0,
      "distillation_loss": 0.022744864225387573,
      "epoch": 7.84,
      "learning_rate": 4.825036940451259e-05,
      "loss": 0.0312,
      "step": 8259,
      "task_loss": 0.10733005404472351
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999248555763,
      "compression_loss": 0.0,
      "distillation_loss": 0.19848661124706268,
      "epoch": 7.84,
      "learning_rate": 4.8246451254250145e-05,
      "loss": 0.1952,
      "step": 8260,
      "task_loss": 0.1654394567012787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999262135749,
      "compression_loss": 0.0,
      "distillation_loss": 0.038717396557331085,
      "epoch": 7.85,
      "learning_rate": 4.8242528881157866e-05,
      "loss": 0.0535,
      "step": 8261,
      "task_loss": 0.18647165596485138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999992755511312,
      "compression_loss": 0.0,
      "distillation_loss": 0.07267004996538162,
      "epoch": 7.85,
      "learning_rate": 4.823860228594829e-05,
      "loss": 0.0775,
      "step": 8262,
      "task_loss": 0.1208515465259552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999992888029132,
      "compression_loss": 0.0,
      "distillation_loss": 0.07859297096729279,
      "epoch": 7.85,
      "learning_rate": 4.823467146933468e-05,
      "loss": 0.0764,
      "step": 8263,
      "task_loss": 0.05707138776779175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999993018920986,
      "compression_loss": 0.0,
      "distillation_loss": 0.015503250993788242,
      "epoch": 7.85,
      "learning_rate": 4.823073643203111e-05,
      "loss": 0.0248,
      "step": 8264,
      "task_loss": 0.10833414644002914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999993148196912,
      "compression_loss": 0.0,
      "distillation_loss": 0.17332197725772858,
      "epoch": 7.85,
      "learning_rate": 4.822679717475237e-05,
      "loss": 0.1677,
      "step": 8265,
      "task_loss": 0.11748655885457993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999993275866947,
      "compression_loss": 0.0,
      "distillation_loss": 0.031857073307037354,
      "epoch": 7.85,
      "learning_rate": 4.8222853698214076e-05,
      "loss": 0.0292,
      "step": 8266,
      "task_loss": 0.005062129348516464
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999993401941128,
      "compression_loss": 0.0,
      "distillation_loss": 0.02804490551352501,
      "epoch": 7.85,
      "learning_rate": 4.8218906003132555e-05,
      "loss": 0.0305,
      "step": 8267,
      "task_loss": 0.05293525010347366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999993526429491,
      "compression_loss": 0.0,
      "distillation_loss": 0.024678591638803482,
      "epoch": 7.85,
      "learning_rate": 4.8214954090224946e-05,
      "loss": 0.0295,
      "step": 8268,
      "task_loss": 0.07294875383377075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999993649342072,
      "compression_loss": 0.0,
      "distillation_loss": 0.08018738776445389,
      "epoch": 7.85,
      "learning_rate": 4.8210997960209114e-05,
      "loss": 0.0856,
      "step": 8269,
      "task_loss": 0.13434135913848877
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999377068891,
      "compression_loss": 0.0,
      "distillation_loss": 0.021488351747393608,
      "epoch": 7.85,
      "learning_rate": 4.8207037613803715e-05,
      "loss": 0.0198,
      "step": 8270,
      "task_loss": 0.00473182275891304
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999389048004,
      "compression_loss": 0.0,
      "distillation_loss": 0.1669892966747284,
      "epoch": 7.85,
      "learning_rate": 4.820307305172818e-05,
      "loss": 0.1599,
      "step": 8271,
      "task_loss": 0.09594349563121796
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999994008725501,
      "compression_loss": 0.0,
      "distillation_loss": 0.039625514298677444,
      "epoch": 7.86,
      "learning_rate": 4.8199104274702666e-05,
      "loss": 0.054,
      "step": 8272,
      "task_loss": 0.18374371528625488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999994125435328,
      "compression_loss": 0.0,
      "distillation_loss": 0.01832488365471363,
      "epoch": 7.86,
      "learning_rate": 4.819513128344814e-05,
      "loss": 0.0221,
      "step": 8273,
      "task_loss": 0.055578745901584625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999994240619557,
      "compression_loss": 0.0,
      "distillation_loss": 0.1487281322479248,
      "epoch": 7.86,
      "learning_rate": 4.8191154078686306e-05,
      "loss": 0.1573,
      "step": 8274,
      "task_loss": 0.23400062322616577
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999994354288227,
      "compression_loss": 0.0,
      "distillation_loss": 0.101641446352005,
      "epoch": 7.86,
      "learning_rate": 4.8187172661139636e-05,
      "loss": 0.1103,
      "step": 8275,
      "task_loss": 0.18786770105361938
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999994466451373,
      "compression_loss": 0.0,
      "distillation_loss": 0.019857624545693398,
      "epoch": 7.86,
      "learning_rate": 4.818318703153139e-05,
      "loss": 0.0187,
      "step": 8276,
      "task_loss": 0.00832618772983551
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999994577119034,
      "compression_loss": 0.0,
      "distillation_loss": 0.034556448459625244,
      "epoch": 7.86,
      "learning_rate": 4.817919719058557e-05,
      "loss": 0.0398,
      "step": 8277,
      "task_loss": 0.08679534494876862
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999994686301245,
      "compression_loss": 0.0,
      "distillation_loss": 0.08879555761814117,
      "epoch": 7.86,
      "learning_rate": 4.8175203139026934e-05,
      "loss": 0.0846,
      "step": 8278,
      "task_loss": 0.046484023332595825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999994794008043,
      "compression_loss": 0.0,
      "distillation_loss": 0.035275399684906006,
      "epoch": 7.86,
      "learning_rate": 4.817120487758104e-05,
      "loss": 0.0353,
      "step": 8279,
      "task_loss": 0.03573717176914215
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999994900249465,
      "compression_loss": 0.0,
      "distillation_loss": 0.05630365014076233,
      "epoch": 7.86,
      "learning_rate": 4.81672024069742e-05,
      "loss": 0.0763,
      "step": 8280,
      "task_loss": 0.2562328577041626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999500503555,
      "compression_loss": 0.0,
      "distillation_loss": 0.03214915469288826,
      "epoch": 7.86,
      "learning_rate": 4.816319572793345e-05,
      "loss": 0.0307,
      "step": 8281,
      "task_loss": 0.01762857846915722
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999510837633,
      "compression_loss": 0.0,
      "distillation_loss": 0.12708397209644318,
      "epoch": 7.87,
      "learning_rate": 4.815918484118665e-05,
      "loss": 0.1273,
      "step": 8282,
      "task_loss": 0.1288326233625412
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999995210281846,
      "compression_loss": 0.0,
      "distillation_loss": 0.03236864507198334,
      "epoch": 7.87,
      "learning_rate": 4.815516974746239e-05,
      "loss": 0.0354,
      "step": 8283,
      "task_loss": 0.06306421756744385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999995310762132,
      "compression_loss": 0.0,
      "distillation_loss": 0.20854994654655457,
      "epoch": 7.87,
      "learning_rate": 4.815115044749003e-05,
      "loss": 0.2032,
      "step": 8284,
      "task_loss": 0.15457791090011597
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999995409827229,
      "compression_loss": 0.0,
      "distillation_loss": 0.10258646309375763,
      "epoch": 7.87,
      "learning_rate": 4.814712694199969e-05,
      "loss": 0.1004,
      "step": 8285,
      "task_loss": 0.08033056557178497
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999995507487169,
      "compression_loss": 0.0,
      "distillation_loss": 0.009589407593011856,
      "epoch": 7.87,
      "learning_rate": 4.814309923172227e-05,
      "loss": 0.0089,
      "step": 8286,
      "task_loss": 0.003172697499394417
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999995603751991,
      "compression_loss": 0.0,
      "distillation_loss": 0.07792666554450989,
      "epoch": 7.87,
      "learning_rate": 4.81390673173894e-05,
      "loss": 0.0795,
      "step": 8287,
      "task_loss": 0.09386920928955078
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999995698631732,
      "compression_loss": 0.0,
      "distillation_loss": 0.19815564155578613,
      "epoch": 7.87,
      "learning_rate": 4.8135031199733524e-05,
      "loss": 0.2015,
      "step": 8288,
      "task_loss": 0.2318621575832367
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999995792136428,
      "compression_loss": 0.0,
      "distillation_loss": 0.06799386441707611,
      "epoch": 7.87,
      "learning_rate": 4.813099087948781e-05,
      "loss": 0.0688,
      "step": 8289,
      "task_loss": 0.07573655992746353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999995884276118,
      "compression_loss": 0.0,
      "distillation_loss": 0.021318640559911728,
      "epoch": 7.87,
      "learning_rate": 4.812694635738621e-05,
      "loss": 0.0198,
      "step": 8290,
      "task_loss": 0.0060688890516757965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999995975060836,
      "compression_loss": 0.0,
      "distillation_loss": 0.06302770227193832,
      "epoch": 7.87,
      "learning_rate": 4.812289763416341e-05,
      "loss": 0.0652,
      "step": 8291,
      "task_loss": 0.08448217064142227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999606450062,
      "compression_loss": 0.0,
      "distillation_loss": 0.07862488925457001,
      "epoch": 7.87,
      "learning_rate": 4.81188447105549e-05,
      "loss": 0.0874,
      "step": 8292,
      "task_loss": 0.16682052612304688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999996152605506,
      "compression_loss": 0.0,
      "distillation_loss": 0.024814866483211517,
      "epoch": 7.88,
      "learning_rate": 4.811478758729691e-05,
      "loss": 0.0228,
      "step": 8293,
      "task_loss": 0.004621252417564392
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999996239385534,
      "compression_loss": 0.0,
      "distillation_loss": 0.04633516073226929,
      "epoch": 7.88,
      "learning_rate": 4.811072626512642e-05,
      "loss": 0.0477,
      "step": 8294,
      "task_loss": 0.06027062237262726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999996324850737,
      "compression_loss": 0.0,
      "distillation_loss": 0.12473642826080322,
      "epoch": 7.88,
      "learning_rate": 4.810666074478121e-05,
      "loss": 0.123,
      "step": 8295,
      "task_loss": 0.1078307181596756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999996409011153,
      "compression_loss": 0.0,
      "distillation_loss": 0.0652468129992485,
      "epoch": 7.88,
      "learning_rate": 4.8102591026999796e-05,
      "loss": 0.0725,
      "step": 8296,
      "task_loss": 0.1382291316986084
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999649187682,
      "compression_loss": 0.0,
      "distillation_loss": 0.011844221502542496,
      "epoch": 7.88,
      "learning_rate": 4.8098517112521456e-05,
      "loss": 0.025,
      "step": 8297,
      "task_loss": 0.1432102769613266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999996573457773,
      "compression_loss": 0.0,
      "distillation_loss": 0.11007647216320038,
      "epoch": 7.88,
      "learning_rate": 4.8094439002086234e-05,
      "loss": 0.1154,
      "step": 8298,
      "task_loss": 0.16329604387283325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999665376405,
      "compression_loss": 0.0,
      "distillation_loss": 0.03129512071609497,
      "epoch": 7.88,
      "learning_rate": 4.809035669643495e-05,
      "loss": 0.0287,
      "step": 8299,
      "task_loss": 0.004948470741510391
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999996732805688,
      "compression_loss": 0.0,
      "distillation_loss": 0.03155820071697235,
      "epoch": 7.88,
      "learning_rate": 4.808627019630917e-05,
      "loss": 0.0389,
      "step": 8300,
      "task_loss": 0.10458790510892868
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999996810592722,
      "compression_loss": 0.0,
      "distillation_loss": 0.018336599692702293,
      "epoch": 7.88,
      "learning_rate": 4.808217950245122e-05,
      "loss": 0.0173,
      "step": 8301,
      "task_loss": 0.008344225585460663
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999996887135191,
      "compression_loss": 0.0,
      "distillation_loss": 0.1453208029270172,
      "epoch": 7.88,
      "learning_rate": 4.807808461560419e-05,
      "loss": 0.153,
      "step": 8302,
      "task_loss": 0.22258791327476501
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999996962443132,
      "compression_loss": 0.0,
      "distillation_loss": 0.026547754183411598,
      "epoch": 7.89,
      "learning_rate": 4.8073985536511956e-05,
      "loss": 0.0299,
      "step": 8303,
      "task_loss": 0.059667013585567474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999997036526579,
      "compression_loss": 0.0,
      "distillation_loss": 0.054472848773002625,
      "epoch": 7.89,
      "learning_rate": 4.806988226591912e-05,
      "loss": 0.0525,
      "step": 8304,
      "task_loss": 0.03490680456161499
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999997109395572,
      "compression_loss": 0.0,
      "distillation_loss": 0.030267203226685524,
      "epoch": 7.89,
      "learning_rate": 4.806577480457106e-05,
      "loss": 0.0307,
      "step": 8305,
      "task_loss": 0.034910961985588074
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999997181060147,
      "compression_loss": 0.0,
      "distillation_loss": 0.08685198426246643,
      "epoch": 7.89,
      "learning_rate": 4.8061663153213935e-05,
      "loss": 0.0862,
      "step": 8306,
      "task_loss": 0.07995598018169403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999725153034,
      "compression_loss": 0.0,
      "distillation_loss": 0.05820311978459358,
      "epoch": 7.89,
      "learning_rate": 4.805754731259462e-05,
      "loss": 0.0552,
      "step": 8307,
      "task_loss": 0.02777511440217495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999997320816187,
      "compression_loss": 0.0,
      "distillation_loss": 0.0553501695394516,
      "epoch": 7.89,
      "learning_rate": 4.805342728346079e-05,
      "loss": 0.0533,
      "step": 8308,
      "task_loss": 0.034611959010362625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999997388927726,
      "compression_loss": 0.0,
      "distillation_loss": 0.06473010778427124,
      "epoch": 7.89,
      "learning_rate": 4.804930306656087e-05,
      "loss": 0.0656,
      "step": 8309,
      "task_loss": 0.07328800857067108
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999997455874995,
      "compression_loss": 0.0,
      "distillation_loss": 0.052684955298900604,
      "epoch": 7.89,
      "learning_rate": 4.804517466264405e-05,
      "loss": 0.055,
      "step": 8310,
      "task_loss": 0.07596210390329361
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999997521668029,
      "compression_loss": 0.0,
      "distillation_loss": 0.1126926839351654,
      "epoch": 7.89,
      "learning_rate": 4.8041042072460244e-05,
      "loss": 0.1187,
      "step": 8311,
      "task_loss": 0.17313697934150696
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999997586316866,
      "compression_loss": 0.0,
      "distillation_loss": 0.2039157748222351,
      "epoch": 7.89,
      "learning_rate": 4.803690529676019e-05,
      "loss": 0.2047,
      "step": 8312,
      "task_loss": 0.211942657828331
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999997649831541,
      "compression_loss": 0.0,
      "distillation_loss": 0.04965321347117424,
      "epoch": 7.89,
      "learning_rate": 4.803276433629534e-05,
      "loss": 0.0641,
      "step": 8313,
      "task_loss": 0.19365498423576355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999997712222093,
      "compression_loss": 0.0,
      "distillation_loss": 0.028864435851573944,
      "epoch": 7.9,
      "learning_rate": 4.802861919181793e-05,
      "loss": 0.0268,
      "step": 8314,
      "task_loss": 0.007797591388225555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999997773498557,
      "compression_loss": 0.0,
      "distillation_loss": 0.06014259159564972,
      "epoch": 7.9,
      "learning_rate": 4.802446986408093e-05,
      "loss": 0.0624,
      "step": 8315,
      "task_loss": 0.0831136405467987
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999997833670972,
      "compression_loss": 0.0,
      "distillation_loss": 0.03688354790210724,
      "epoch": 7.9,
      "learning_rate": 4.8020316353838095e-05,
      "loss": 0.0529,
      "step": 8316,
      "task_loss": 0.19723045825958252
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999997892749372,
      "compression_loss": 0.0,
      "distillation_loss": 0.029626084491610527,
      "epoch": 7.9,
      "learning_rate": 4.8016158661843926e-05,
      "loss": 0.0273,
      "step": 8317,
      "task_loss": 0.006546778604388237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999997950743797,
      "compression_loss": 0.0,
      "distillation_loss": 0.02643163688480854,
      "epoch": 7.9,
      "learning_rate": 4.8011996788853686e-05,
      "loss": 0.0339,
      "step": 8318,
      "task_loss": 0.10074707865715027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998007664281,
      "compression_loss": 0.0,
      "distillation_loss": 0.021654170006513596,
      "epoch": 7.9,
      "learning_rate": 4.80078307356234e-05,
      "loss": 0.0306,
      "step": 8319,
      "task_loss": 0.1115964949131012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998063520862,
      "compression_loss": 0.0,
      "distillation_loss": 0.020197220146656036,
      "epoch": 7.9,
      "learning_rate": 4.800366050290986e-05,
      "loss": 0.0187,
      "step": 8320,
      "task_loss": 0.005556017160415649
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998118323577,
      "compression_loss": 0.0,
      "distillation_loss": 0.015532903373241425,
      "epoch": 7.9,
      "learning_rate": 4.799948609147061e-05,
      "loss": 0.0234,
      "step": 8321,
      "task_loss": 0.09371798485517502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998172082463,
      "compression_loss": 0.0,
      "distillation_loss": 0.041199732571840286,
      "epoch": 7.9,
      "learning_rate": 4.7995307502063936e-05,
      "loss": 0.0383,
      "step": 8322,
      "task_loss": 0.011793764308094978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998224807556,
      "compression_loss": 0.0,
      "distillation_loss": 0.11037556827068329,
      "epoch": 7.9,
      "learning_rate": 4.799112473544891e-05,
      "loss": 0.1096,
      "step": 8323,
      "task_loss": 0.1021459624171257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998276508894,
      "compression_loss": 0.0,
      "distillation_loss": 0.05237120762467384,
      "epoch": 7.91,
      "learning_rate": 4.7986937792385344e-05,
      "loss": 0.0535,
      "step": 8324,
      "task_loss": 0.0640583410859108
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998327196512,
      "compression_loss": 0.0,
      "distillation_loss": 0.0475940927863121,
      "epoch": 7.91,
      "learning_rate": 4.798274667363383e-05,
      "loss": 0.0521,
      "step": 8325,
      "task_loss": 0.09266771376132965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998376880448,
      "compression_loss": 0.0,
      "distillation_loss": 0.022809145972132683,
      "epoch": 7.91,
      "learning_rate": 4.7978551379955684e-05,
      "loss": 0.0212,
      "step": 8326,
      "task_loss": 0.006463898345828056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999842557074,
      "compression_loss": 0.0,
      "distillation_loss": 0.11664307862520218,
      "epoch": 7.91,
      "learning_rate": 4.797435191211302e-05,
      "loss": 0.1186,
      "step": 8327,
      "task_loss": 0.13656139373779297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998473277422,
      "compression_loss": 0.0,
      "distillation_loss": 0.047932952642440796,
      "epoch": 7.91,
      "learning_rate": 4.797014827086869e-05,
      "loss": 0.0526,
      "step": 8328,
      "task_loss": 0.09455284476280212
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998520010533,
      "compression_loss": 0.0,
      "distillation_loss": 0.0719044953584671,
      "epoch": 7.91,
      "learning_rate": 4.79659404569863e-05,
      "loss": 0.0775,
      "step": 8329,
      "task_loss": 0.1278771162033081
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999856578011,
      "compression_loss": 0.0,
      "distillation_loss": 0.010797183960676193,
      "epoch": 7.91,
      "learning_rate": 4.7961728471230214e-05,
      "loss": 0.0099,
      "step": 8330,
      "task_loss": 0.0018469560891389847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998610596187,
      "compression_loss": 0.0,
      "distillation_loss": 0.0197126641869545,
      "epoch": 7.91,
      "learning_rate": 4.7957512314365574e-05,
      "loss": 0.0256,
      "step": 8331,
      "task_loss": 0.07906901836395264
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998654468805,
      "compression_loss": 0.0,
      "distillation_loss": 0.02190200798213482,
      "epoch": 7.91,
      "learning_rate": 4.7953291987158254e-05,
      "loss": 0.0201,
      "step": 8332,
      "task_loss": 0.003983369097113609
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998697407997,
      "compression_loss": 0.0,
      "distillation_loss": 0.15597021579742432,
      "epoch": 7.91,
      "learning_rate": 4.79490674903749e-05,
      "loss": 0.1605,
      "step": 8333,
      "task_loss": 0.20110704004764557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998739423803,
      "compression_loss": 0.0,
      "distillation_loss": 0.020552337169647217,
      "epoch": 7.91,
      "learning_rate": 4.7944838824782916e-05,
      "loss": 0.0292,
      "step": 8334,
      "task_loss": 0.10749036073684692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998780526257,
      "compression_loss": 0.0,
      "distillation_loss": 0.02331959456205368,
      "epoch": 7.92,
      "learning_rate": 4.794060599115045e-05,
      "loss": 0.0217,
      "step": 8335,
      "task_loss": 0.0067675188183784485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998820725398,
      "compression_loss": 0.0,
      "distillation_loss": 0.012297404929995537,
      "epoch": 7.92,
      "learning_rate": 4.793636899024643e-05,
      "loss": 0.019,
      "step": 8336,
      "task_loss": 0.07974053174257278
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998860031262,
      "compression_loss": 0.0,
      "distillation_loss": 0.17458131909370422,
      "epoch": 7.92,
      "learning_rate": 4.7932127822840516e-05,
      "loss": 0.1665,
      "step": 8337,
      "task_loss": 0.09329235553741455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998898453885,
      "compression_loss": 0.0,
      "distillation_loss": 0.016357656568288803,
      "epoch": 7.92,
      "learning_rate": 4.792788248970314e-05,
      "loss": 0.0217,
      "step": 8338,
      "task_loss": 0.06964881718158722
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998936003305,
      "compression_loss": 0.0,
      "distillation_loss": 0.10674792528152466,
      "epoch": 7.92,
      "learning_rate": 4.79236329916055e-05,
      "loss": 0.1097,
      "step": 8339,
      "task_loss": 0.13648146390914917
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999998972689558,
      "compression_loss": 0.0,
      "distillation_loss": 0.014713255688548088,
      "epoch": 7.92,
      "learning_rate": 4.79193793293195e-05,
      "loss": 0.0227,
      "step": 8340,
      "task_loss": 0.09422945976257324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999008522681,
      "compression_loss": 0.0,
      "distillation_loss": 0.014077425003051758,
      "epoch": 7.92,
      "learning_rate": 4.791512150361788e-05,
      "loss": 0.0292,
      "step": 8341,
      "task_loss": 0.1649925708770752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999043512712,
      "compression_loss": 0.0,
      "distillation_loss": 0.08245790749788284,
      "epoch": 7.92,
      "learning_rate": 4.791085951527408e-05,
      "loss": 0.0749,
      "step": 8342,
      "task_loss": 0.006448717787861824
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999077669687,
      "compression_loss": 0.0,
      "distillation_loss": 0.08674632757902145,
      "epoch": 7.92,
      "learning_rate": 4.7906593365062304e-05,
      "loss": 0.0895,
      "step": 8343,
      "task_loss": 0.11430380493402481
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999111003642,
      "compression_loss": 0.0,
      "distillation_loss": 0.06291348487138748,
      "epoch": 7.92,
      "learning_rate": 4.790232305375752e-05,
      "loss": 0.061,
      "step": 8344,
      "task_loss": 0.04396482929587364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999143524614,
      "compression_loss": 0.0,
      "distillation_loss": 0.07400640845298767,
      "epoch": 7.92,
      "learning_rate": 4.789804858213547e-05,
      "loss": 0.0753,
      "step": 8345,
      "task_loss": 0.08731956779956818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999175242641,
      "compression_loss": 0.0,
      "distillation_loss": 0.014547398313879967,
      "epoch": 7.93,
      "learning_rate": 4.7893769950972605e-05,
      "loss": 0.0196,
      "step": 8346,
      "task_loss": 0.06489875912666321
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999206167759,
      "compression_loss": 0.0,
      "distillation_loss": 0.02471366710960865,
      "epoch": 7.93,
      "learning_rate": 4.788948716104618e-05,
      "loss": 0.0266,
      "step": 8347,
      "task_loss": 0.043872442096471786
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999236310006,
      "compression_loss": 0.0,
      "distillation_loss": 0.032343026250600815,
      "epoch": 7.93,
      "learning_rate": 4.7885200213134164e-05,
      "loss": 0.0309,
      "step": 8348,
      "task_loss": 0.01805000938475132
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999265679416,
      "compression_loss": 0.0,
      "distillation_loss": 0.16060638427734375,
      "epoch": 7.93,
      "learning_rate": 4.788090910801532e-05,
      "loss": 0.1635,
      "step": 8349,
      "task_loss": 0.18905338644981384
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999294286029,
      "compression_loss": 0.0,
      "distillation_loss": 0.021082771942019463,
      "epoch": 7.93,
      "learning_rate": 4.787661384646913e-05,
      "loss": 0.0198,
      "step": 8350,
      "task_loss": 0.008079813793301582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999932213988,
      "compression_loss": 0.0,
      "distillation_loss": 0.09513162821531296,
      "epoch": 7.93,
      "learning_rate": 4.787231442927587e-05,
      "loss": 0.1033,
      "step": 8351,
      "task_loss": 0.17651014029979706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999349251006,
      "compression_loss": 0.0,
      "distillation_loss": 0.026541750878095627,
      "epoch": 7.93,
      "learning_rate": 4.786801085721654e-05,
      "loss": 0.0317,
      "step": 8352,
      "task_loss": 0.07830867916345596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999375629444,
      "compression_loss": 0.0,
      "distillation_loss": 0.052206460386514664,
      "epoch": 7.93,
      "learning_rate": 4.78637031310729e-05,
      "loss": 0.0684,
      "step": 8353,
      "task_loss": 0.21389958262443542
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999401285232,
      "compression_loss": 0.0,
      "distillation_loss": 0.020280838012695312,
      "epoch": 7.93,
      "learning_rate": 4.7859391251627474e-05,
      "loss": 0.0348,
      "step": 8354,
      "task_loss": 0.16505402326583862
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999426228404,
      "compression_loss": 0.0,
      "distillation_loss": 0.04192467778921127,
      "epoch": 7.93,
      "learning_rate": 4.7855075219663535e-05,
      "loss": 0.0557,
      "step": 8355,
      "task_loss": 0.17976054549217224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999450469,
      "compression_loss": 0.0,
      "distillation_loss": 0.029960088431835175,
      "epoch": 7.94,
      "learning_rate": 4.785075503596511e-05,
      "loss": 0.0296,
      "step": 8356,
      "task_loss": 0.02586023323237896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999474017054,
      "compression_loss": 0.0,
      "distillation_loss": 0.041166506707668304,
      "epoch": 7.94,
      "learning_rate": 4.7846430701316994e-05,
      "loss": 0.0383,
      "step": 8357,
      "task_loss": 0.012875651940703392
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999496882605,
      "compression_loss": 0.0,
      "distillation_loss": 0.07181466370820999,
      "epoch": 7.94,
      "learning_rate": 4.78421022165047e-05,
      "loss": 0.0763,
      "step": 8358,
      "task_loss": 0.11669700592756271
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999519075689,
      "compression_loss": 0.0,
      "distillation_loss": 0.1549142599105835,
      "epoch": 7.94,
      "learning_rate": 4.783776958231453e-05,
      "loss": 0.1549,
      "step": 8359,
      "task_loss": 0.1552504003047943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999540606343,
      "compression_loss": 0.0,
      "distillation_loss": 0.0709429532289505,
      "epoch": 7.94,
      "learning_rate": 4.783343279953353e-05,
      "loss": 0.0802,
      "step": 8360,
      "task_loss": 0.16332530975341797
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999561484603,
      "compression_loss": 0.0,
      "distillation_loss": 0.05141635984182358,
      "epoch": 7.94,
      "learning_rate": 4.782909186894949e-05,
      "loss": 0.048,
      "step": 8361,
      "task_loss": 0.017139893025159836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999581720506,
      "compression_loss": 0.0,
      "distillation_loss": 0.08039192855358124,
      "epoch": 7.94,
      "learning_rate": 4.782474679135097e-05,
      "loss": 0.0835,
      "step": 8362,
      "task_loss": 0.11189291626214981
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999960132409,
      "compression_loss": 0.0,
      "distillation_loss": 0.03790656104683876,
      "epoch": 7.94,
      "learning_rate": 4.782039756752727e-05,
      "loss": 0.0349,
      "step": 8363,
      "task_loss": 0.008079200983047485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999620305391,
      "compression_loss": 0.0,
      "distillation_loss": 0.03653489425778389,
      "epoch": 7.94,
      "learning_rate": 4.781604419826845e-05,
      "loss": 0.043,
      "step": 8364,
      "task_loss": 0.10079368948936462
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999638674447,
      "compression_loss": 0.0,
      "distillation_loss": 0.06203051656484604,
      "epoch": 7.94,
      "learning_rate": 4.781168668436532e-05,
      "loss": 0.0605,
      "step": 8365,
      "task_loss": 0.04713946580886841
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999656441292,
      "compression_loss": 0.0,
      "distillation_loss": 0.1302792727947235,
      "epoch": 7.94,
      "learning_rate": 4.780732502660943e-05,
      "loss": 0.1366,
      "step": 8366,
      "task_loss": 0.1930633783340454
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999673615966,
      "compression_loss": 0.0,
      "distillation_loss": 0.1713554859161377,
      "epoch": 7.95,
      "learning_rate": 4.780295922579312e-05,
      "loss": 0.1854,
      "step": 8367,
      "task_loss": 0.3117556571960449
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999690208504,
      "compression_loss": 0.0,
      "distillation_loss": 0.08704891055822372,
      "epoch": 7.95,
      "learning_rate": 4.779858928270944e-05,
      "loss": 0.0917,
      "step": 8368,
      "task_loss": 0.13358929753303528
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999706228943,
      "compression_loss": 0.0,
      "distillation_loss": 0.12303698807954788,
      "epoch": 7.95,
      "learning_rate": 4.7794215198152216e-05,
      "loss": 0.1175,
      "step": 8369,
      "task_loss": 0.06791896373033524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999972168732,
      "compression_loss": 0.0,
      "distillation_loss": 0.09498666226863861,
      "epoch": 7.95,
      "learning_rate": 4.778983697291603e-05,
      "loss": 0.091,
      "step": 8370,
      "task_loss": 0.05467312037944794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999736593671,
      "compression_loss": 0.0,
      "distillation_loss": 0.0775984525680542,
      "epoch": 7.95,
      "learning_rate": 4.7785454607796195e-05,
      "loss": 0.0746,
      "step": 8371,
      "task_loss": 0.04722478240728378
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999750958036,
      "compression_loss": 0.0,
      "distillation_loss": 0.09941750019788742,
      "epoch": 7.95,
      "learning_rate": 4.77810681035888e-05,
      "loss": 0.0948,
      "step": 8372,
      "task_loss": 0.053179264068603516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999764790447,
      "compression_loss": 0.0,
      "distillation_loss": 0.08102773874998093,
      "epoch": 7.95,
      "learning_rate": 4.777667746109067e-05,
      "loss": 0.0827,
      "step": 8373,
      "task_loss": 0.09824100136756897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999778100945,
      "compression_loss": 0.0,
      "distillation_loss": 0.06384574621915817,
      "epoch": 7.95,
      "learning_rate": 4.7772282681099377e-05,
      "loss": 0.0756,
      "step": 8374,
      "task_loss": 0.18099090456962585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999790899565,
      "compression_loss": 0.0,
      "distillation_loss": 0.024305138736963272,
      "epoch": 7.95,
      "learning_rate": 4.7767883764413266e-05,
      "loss": 0.0307,
      "step": 8375,
      "task_loss": 0.08778215944766998
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999803196343,
      "compression_loss": 0.0,
      "distillation_loss": 0.029779810458421707,
      "epoch": 7.95,
      "learning_rate": 4.776348071183142e-05,
      "loss": 0.0478,
      "step": 8376,
      "task_loss": 0.20959466695785522
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999815001317,
      "compression_loss": 0.0,
      "distillation_loss": 0.06962715834379196,
      "epoch": 7.96,
      "learning_rate": 4.775907352415367e-05,
      "loss": 0.0651,
      "step": 8377,
      "task_loss": 0.0247165709733963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999826324524,
      "compression_loss": 0.0,
      "distillation_loss": 0.14432743191719055,
      "epoch": 7.96,
      "learning_rate": 4.7754662202180606e-05,
      "loss": 0.1392,
      "step": 8378,
      "task_loss": 0.0935506820678711
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999837176001,
      "compression_loss": 0.0,
      "distillation_loss": 0.019188987091183662,
      "epoch": 7.96,
      "learning_rate": 4.7750246746713565e-05,
      "loss": 0.0178,
      "step": 8379,
      "task_loss": 0.005380744114518166
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999847565783,
      "compression_loss": 0.0,
      "distillation_loss": 0.04302148520946503,
      "epoch": 7.96,
      "learning_rate": 4.7745827158554634e-05,
      "loss": 0.071,
      "step": 8380,
      "task_loss": 0.3228791356086731
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999857503908,
      "compression_loss": 0.0,
      "distillation_loss": 0.033299144357442856,
      "epoch": 7.96,
      "learning_rate": 4.774140343850666e-05,
      "loss": 0.0338,
      "step": 8381,
      "task_loss": 0.03854818642139435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999867000414,
      "compression_loss": 0.0,
      "distillation_loss": 0.16388925909996033,
      "epoch": 7.96,
      "learning_rate": 4.773697558737322e-05,
      "loss": 0.1616,
      "step": 8382,
      "task_loss": 0.14093096554279327
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999876065336,
      "compression_loss": 0.0,
      "distillation_loss": 0.11715799570083618,
      "epoch": 7.96,
      "learning_rate": 4.773254360595867e-05,
      "loss": 0.1141,
      "step": 8383,
      "task_loss": 0.08705883473157883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999884708712,
      "compression_loss": 0.0,
      "distillation_loss": 0.07380795478820801,
      "epoch": 7.96,
      "learning_rate": 4.77281074950681e-05,
      "loss": 0.0678,
      "step": 8384,
      "task_loss": 0.013948189094662666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999892940577,
      "compression_loss": 0.0,
      "distillation_loss": 0.15698346495628357,
      "epoch": 7.96,
      "learning_rate": 4.7723667255507334e-05,
      "loss": 0.1518,
      "step": 8385,
      "task_loss": 0.10555059462785721
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999990077097,
      "compression_loss": 0.0,
      "distillation_loss": 0.02515373006463051,
      "epoch": 7.96,
      "learning_rate": 4.771922288808297e-05,
      "loss": 0.0234,
      "step": 8386,
      "task_loss": 0.007526658475399017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999908209927,
      "compression_loss": 0.0,
      "distillation_loss": 0.015803663060069084,
      "epoch": 7.96,
      "learning_rate": 4.771477439360235e-05,
      "loss": 0.0145,
      "step": 8387,
      "task_loss": 0.002915032207965851
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999915267485,
      "compression_loss": 0.0,
      "distillation_loss": 0.12163722515106201,
      "epoch": 7.97,
      "learning_rate": 4.7710321772873566e-05,
      "loss": 0.1286,
      "step": 8388,
      "task_loss": 0.19121067225933075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999921953681,
      "compression_loss": 0.0,
      "distillation_loss": 0.04631570726633072,
      "epoch": 7.97,
      "learning_rate": 4.770586502670546e-05,
      "loss": 0.054,
      "step": 8389,
      "task_loss": 0.12329075485467911
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999992827855,
      "compression_loss": 0.0,
      "distillation_loss": 0.15134918689727783,
      "epoch": 7.97,
      "learning_rate": 4.770140415590762e-05,
      "loss": 0.1518,
      "step": 8390,
      "task_loss": 0.15575477480888367
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999934252132,
      "compression_loss": 0.0,
      "distillation_loss": 0.07986609637737274,
      "epoch": 7.97,
      "learning_rate": 4.769693916129039e-05,
      "loss": 0.0843,
      "step": 8391,
      "task_loss": 0.12397737801074982
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999939884461,
      "compression_loss": 0.0,
      "distillation_loss": 0.09135753661394119,
      "epoch": 7.97,
      "learning_rate": 4.769247004366485e-05,
      "loss": 0.0998,
      "step": 8392,
      "task_loss": 0.1758865863084793
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999945185575,
      "compression_loss": 0.0,
      "distillation_loss": 0.029954630881547928,
      "epoch": 7.97,
      "learning_rate": 4.768799680384283e-05,
      "loss": 0.0286,
      "step": 8393,
      "task_loss": 0.016476107761263847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999950165512,
      "compression_loss": 0.0,
      "distillation_loss": 0.1209501326084137,
      "epoch": 7.97,
      "learning_rate": 4.768351944263693e-05,
      "loss": 0.1188,
      "step": 8394,
      "task_loss": 0.09983272105455399
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999954834306,
      "compression_loss": 0.0,
      "distillation_loss": 0.12946045398712158,
      "epoch": 7.97,
      "learning_rate": 4.767903796086048e-05,
      "loss": 0.1457,
      "step": 8395,
      "task_loss": 0.29160839319229126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999959201997,
      "compression_loss": 0.0,
      "distillation_loss": 0.03893984854221344,
      "epoch": 7.97,
      "learning_rate": 4.767455235932756e-05,
      "loss": 0.0484,
      "step": 8396,
      "task_loss": 0.13344644010066986
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999963278618,
      "compression_loss": 0.0,
      "distillation_loss": 0.05474621057510376,
      "epoch": 7.97,
      "learning_rate": 4.7670062638853e-05,
      "loss": 0.0528,
      "step": 8397,
      "task_loss": 0.03571630269289017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999996707421,
      "compression_loss": 0.0,
      "distillation_loss": 0.05862941965460777,
      "epoch": 7.98,
      "learning_rate": 4.766556880025238e-05,
      "loss": 0.0592,
      "step": 8398,
      "task_loss": 0.06433902680873871
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999970598807,
      "compression_loss": 0.0,
      "distillation_loss": 0.022588767111301422,
      "epoch": 7.98,
      "learning_rate": 4.7661070844342033e-05,
      "loss": 0.0212,
      "step": 8399,
      "task_loss": 0.008752534165978432
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999973862446,
      "compression_loss": 0.0,
      "distillation_loss": 0.0467553474009037,
      "epoch": 7.98,
      "learning_rate": 4.7656568771939024e-05,
      "loss": 0.0514,
      "step": 8400,
      "task_loss": 0.09337884187698364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999976875165,
      "compression_loss": 0.0,
      "distillation_loss": 0.11045566201210022,
      "epoch": 7.98,
      "learning_rate": 4.765206258386119e-05,
      "loss": 0.1095,
      "step": 8401,
      "task_loss": 0.10068619251251221
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999979647,
      "compression_loss": 0.0,
      "distillation_loss": 0.022844959050416946,
      "epoch": 7.98,
      "learning_rate": 4.7647552280927086e-05,
      "loss": 0.0213,
      "step": 8402,
      "task_loss": 0.007824547588825226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999982187989,
      "compression_loss": 0.0,
      "distillation_loss": 0.038698434829711914,
      "epoch": 7.98,
      "learning_rate": 4.764303786395604e-05,
      "loss": 0.0406,
      "step": 8403,
      "task_loss": 0.05789912864565849
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999984508167,
      "compression_loss": 0.0,
      "distillation_loss": 0.043944694101810455,
      "epoch": 7.98,
      "learning_rate": 4.763851933376812e-05,
      "loss": 0.0417,
      "step": 8404,
      "task_loss": 0.02184183895587921
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999986617573,
      "compression_loss": 0.0,
      "distillation_loss": 0.012151426635682583,
      "epoch": 7.98,
      "learning_rate": 4.763399669118414e-05,
      "loss": 0.0115,
      "step": 8405,
      "task_loss": 0.005970221012830734
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999988526242,
      "compression_loss": 0.0,
      "distillation_loss": 0.0856751874089241,
      "epoch": 7.98,
      "learning_rate": 4.762946993702565e-05,
      "loss": 0.0912,
      "step": 8406,
      "task_loss": 0.1414024531841278
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999990244211,
      "compression_loss": 0.0,
      "distillation_loss": 0.059896957129240036,
      "epoch": 7.98,
      "learning_rate": 4.7624939072114954e-05,
      "loss": 0.0598,
      "step": 8407,
      "task_loss": 0.05849459767341614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999991781517,
      "compression_loss": 0.0,
      "distillation_loss": 0.1813950538635254,
      "epoch": 7.98,
      "learning_rate": 4.762040409727512e-05,
      "loss": 0.1745,
      "step": 8408,
      "task_loss": 0.11278204619884491
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999993148197,
      "compression_loss": 0.0,
      "distillation_loss": 0.11629247665405273,
      "epoch": 7.99,
      "learning_rate": 4.761586501332994e-05,
      "loss": 0.1225,
      "step": 8409,
      "task_loss": 0.17883270978927612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999994354289,
      "compression_loss": 0.0,
      "distillation_loss": 0.019268091768026352,
      "epoch": 7.99,
      "learning_rate": 4.7611321821103954e-05,
      "loss": 0.0178,
      "step": 8410,
      "task_loss": 0.0050684306770563126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999995409828,
      "compression_loss": 0.0,
      "distillation_loss": 0.03275691345334053,
      "epoch": 7.99,
      "learning_rate": 4.760677452142247e-05,
      "loss": 0.0434,
      "step": 8411,
      "task_loss": 0.13962477445602417
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999996324851,
      "compression_loss": 0.0,
      "distillation_loss": 0.0782676413655281,
      "epoch": 7.99,
      "learning_rate": 4.760222311511152e-05,
      "loss": 0.0755,
      "step": 8412,
      "task_loss": 0.05032962188124657
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999997109396,
      "compression_loss": 0.0,
      "distillation_loss": 0.03344403952360153,
      "epoch": 7.99,
      "learning_rate": 4.759766760299788e-05,
      "loss": 0.0368,
      "step": 8413,
      "task_loss": 0.06691578030586243
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999997773499,
      "compression_loss": 0.0,
      "distillation_loss": 0.027902770787477493,
      "epoch": 7.99,
      "learning_rate": 4.759310798590909e-05,
      "loss": 0.0359,
      "step": 8414,
      "task_loss": 0.10809982568025589
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999998327197,
      "compression_loss": 0.0,
      "distillation_loss": 0.0703706368803978,
      "epoch": 7.99,
      "learning_rate": 4.758854426467343e-05,
      "loss": 0.0753,
      "step": 8415,
      "task_loss": 0.11971607804298401
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999998780527,
      "compression_loss": 0.0,
      "distillation_loss": 0.06848819553852081,
      "epoch": 7.99,
      "learning_rate": 4.758397644011992e-05,
      "loss": 0.0691,
      "step": 8416,
      "task_loss": 0.07492919266223907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999999143526,
      "compression_loss": 0.0,
      "distillation_loss": 0.030622560530900955,
      "epoch": 7.99,
      "learning_rate": 4.757940451307831e-05,
      "loss": 0.0384,
      "step": 8417,
      "task_loss": 0.10844360291957855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999999426228,
      "compression_loss": 0.0,
      "distillation_loss": 0.06612611562013626,
      "epoch": 7.99,
      "learning_rate": 4.757482848437914e-05,
      "loss": 0.0736,
      "step": 8418,
      "task_loss": 0.14050063490867615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999999638675,
      "compression_loss": 0.0,
      "distillation_loss": 0.0231521874666214,
      "epoch": 8.0,
      "learning_rate": 4.7570248354853644e-05,
      "loss": 0.0454,
      "step": 8419,
      "task_loss": 0.2456911951303482
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.79999999997909,
      "compression_loss": 0.0,
      "distillation_loss": 0.027434296905994415,
      "epoch": 8.0,
      "learning_rate": 4.7565664125333845e-05,
      "loss": 0.0334,
      "step": 8420,
      "task_loss": 0.08753321319818497
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.799999999989294,
      "compression_loss": 0.0,
      "distillation_loss": 0.07552851736545563,
      "epoch": 8.0,
      "learning_rate": 4.7561075796652464e-05,
      "loss": 0.0764,
      "step": 8421,
      "task_loss": 0.08394578844308853
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999999954834,
      "compression_loss": 0.0,
      "distillation_loss": 0.02573383040726185,
      "epoch": 8.0,
      "learning_rate": 4.755648336964302e-05,
      "loss": 0.0239,
      "step": 8422,
      "task_loss": 0.0073777977377176285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999999986618,
      "compression_loss": 0.0,
      "distillation_loss": 0.06676770746707916,
      "epoch": 8.0,
      "learning_rate": 4.7551886845139743e-05,
      "loss": 0.089,
      "step": 8423,
      "task_loss": 0.28941696882247925
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6247876100877703,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.7998046816538257,
      "compression/magnitude_sparsity/target_sparsity_level": 0.7999999999998327,
      "compression_loss": 0.0,
      "distillation_loss": 0.007792739663273096,
      "epoch": 8.0,
      "learning_rate": 4.754728622397761e-05,
      "loss": 0.0073,
      "step": 8424,
      "task_loss": 0.0028934250585734844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1277932971715927,
      "epoch": 8.0,
      "learning_rate": 4.754268150699234e-05,
      "loss": 0.12,
      "step": 8425,
      "task_loss": 0.049523890018463135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.18691036105155945,
      "epoch": 8.0,
      "learning_rate": 4.753807269502041e-05,
      "loss": 0.1797,
      "step": 8426,
      "task_loss": 0.11476030945777893
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11132822930812836,
      "epoch": 8.0,
      "learning_rate": 4.7533459788899026e-05,
      "loss": 0.1066,
      "step": 8427,
      "task_loss": 0.0645451620221138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06175380200147629,
      "epoch": 8.0,
      "learning_rate": 4.752884278946614e-05,
      "loss": 0.0701,
      "step": 8428,
      "task_loss": 0.14483240246772766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10747231543064117,
      "epoch": 8.0,
      "learning_rate": 4.752422169756048e-05,
      "loss": 0.1015,
      "step": 8429,
      "task_loss": 0.0479457788169384
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.15526169538497925,
      "epoch": 8.01,
      "learning_rate": 4.7519596514021464e-05,
      "loss": 0.1475,
      "step": 8430,
      "task_loss": 0.07779216766357422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12560060620307922,
      "epoch": 8.01,
      "learning_rate": 4.751496723968929e-05,
      "loss": 0.1216,
      "step": 8431,
      "task_loss": 0.08531402051448822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06614533811807632,
      "epoch": 8.01,
      "learning_rate": 4.751033387540488e-05,
      "loss": 0.0747,
      "step": 8432,
      "task_loss": 0.151977077126503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.19305379688739777,
      "epoch": 8.01,
      "learning_rate": 4.7505696422009904e-05,
      "loss": 0.1905,
      "step": 8433,
      "task_loss": 0.16793277859687805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05545946955680847,
      "epoch": 8.01,
      "learning_rate": 4.750105488034679e-05,
      "loss": 0.0656,
      "step": 8434,
      "task_loss": 0.15694250166416168
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09607134759426117,
      "epoch": 8.01,
      "learning_rate": 4.749640925125869e-05,
      "loss": 0.1036,
      "step": 8435,
      "task_loss": 0.17179882526397705
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.036613885313272476,
      "epoch": 8.01,
      "learning_rate": 4.749175953558951e-05,
      "loss": 0.0342,
      "step": 8436,
      "task_loss": 0.012063302099704742
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1419782042503357,
      "epoch": 8.01,
      "learning_rate": 4.748710573418388e-05,
      "loss": 0.1527,
      "step": 8437,
      "task_loss": 0.24885791540145874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016515720635652542,
      "epoch": 8.01,
      "learning_rate": 4.7482447847887204e-05,
      "loss": 0.0154,
      "step": 8438,
      "task_loss": 0.00573185458779335
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017054930329322815,
      "epoch": 8.01,
      "learning_rate": 4.747778587754559e-05,
      "loss": 0.0218,
      "step": 8439,
      "task_loss": 0.06488415598869324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03005680814385414,
      "epoch": 8.02,
      "learning_rate": 4.7473119824005926e-05,
      "loss": 0.0275,
      "step": 8440,
      "task_loss": 0.004504000768065453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05955897271633148,
      "epoch": 8.02,
      "learning_rate": 4.7468449688115806e-05,
      "loss": 0.0621,
      "step": 8441,
      "task_loss": 0.08468262106180191
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03443329036235809,
      "epoch": 8.02,
      "learning_rate": 4.74637754707236e-05,
      "loss": 0.0326,
      "step": 8442,
      "task_loss": 0.016446424648165703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020430468022823334,
      "epoch": 8.02,
      "learning_rate": 4.7459097172678386e-05,
      "loss": 0.0197,
      "step": 8443,
      "task_loss": 0.013592688366770744
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12642060220241547,
      "epoch": 8.02,
      "learning_rate": 4.745441479483001e-05,
      "loss": 0.1201,
      "step": 8444,
      "task_loss": 0.06335929036140442
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06663849949836731,
      "epoch": 8.02,
      "learning_rate": 4.744972833802904e-05,
      "loss": 0.0631,
      "step": 8445,
      "task_loss": 0.031001247465610504
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.3165062665939331,
      "epoch": 8.02,
      "learning_rate": 4.74450378031268e-05,
      "loss": 0.3055,
      "step": 8446,
      "task_loss": 0.20671746134757996
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07666581869125366,
      "epoch": 8.02,
      "learning_rate": 4.744034319097535e-05,
      "loss": 0.0726,
      "step": 8447,
      "task_loss": 0.035642966628074646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.050358712673187256,
      "epoch": 8.02,
      "learning_rate": 4.743564450242749e-05,
      "loss": 0.0464,
      "step": 8448,
      "task_loss": 0.010279197245836258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017902474850416183,
      "epoch": 8.02,
      "learning_rate": 4.7430941738336745e-05,
      "loss": 0.0164,
      "step": 8449,
      "task_loss": 0.002874387428164482
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022135000675916672,
      "epoch": 8.02,
      "learning_rate": 4.742623489955741e-05,
      "loss": 0.0304,
      "step": 8450,
      "task_loss": 0.10519418120384216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022276829928159714,
      "epoch": 8.03,
      "learning_rate": 4.74215239869445e-05,
      "loss": 0.0206,
      "step": 8451,
      "task_loss": 0.005232140421867371
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07043208181858063,
      "epoch": 8.03,
      "learning_rate": 4.741680900135377e-05,
      "loss": 0.0705,
      "step": 8452,
      "task_loss": 0.07124508917331696
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0691102147102356,
      "epoch": 8.03,
      "learning_rate": 4.741208994364173e-05,
      "loss": 0.0644,
      "step": 8453,
      "task_loss": 0.021717606112360954
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026366369798779488,
      "epoch": 8.03,
      "learning_rate": 4.740736681466561e-05,
      "loss": 0.0242,
      "step": 8454,
      "task_loss": 0.00459631159901619
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1180252730846405,
      "epoch": 8.03,
      "learning_rate": 4.74026396152834e-05,
      "loss": 0.119,
      "step": 8455,
      "task_loss": 0.1280374825000763
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07965636998414993,
      "epoch": 8.03,
      "learning_rate": 4.7397908346353796e-05,
      "loss": 0.0733,
      "step": 8456,
      "task_loss": 0.01603274792432785
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09866228699684143,
      "epoch": 8.03,
      "learning_rate": 4.739317300873628e-05,
      "loss": 0.0939,
      "step": 8457,
      "task_loss": 0.05102086067199707
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05824364721775055,
      "epoch": 8.03,
      "learning_rate": 4.738843360329104e-05,
      "loss": 0.0556,
      "step": 8458,
      "task_loss": 0.03139631822705269
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0658094584941864,
      "epoch": 8.03,
      "learning_rate": 4.738369013087902e-05,
      "loss": 0.0644,
      "step": 8459,
      "task_loss": 0.05181333050131798
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022087426856160164,
      "epoch": 8.03,
      "learning_rate": 4.7378942592361876e-05,
      "loss": 0.0291,
      "step": 8460,
      "task_loss": 0.09194046258926392
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013795844279229641,
      "epoch": 8.04,
      "learning_rate": 4.737419098860204e-05,
      "loss": 0.0216,
      "step": 8461,
      "task_loss": 0.09192723780870438
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13639725744724274,
      "epoch": 8.04,
      "learning_rate": 4.7369435320462654e-05,
      "loss": 0.1467,
      "step": 8462,
      "task_loss": 0.23894120752811432
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07075132429599762,
      "epoch": 8.04,
      "learning_rate": 4.73646755888076e-05,
      "loss": 0.0742,
      "step": 8463,
      "task_loss": 0.10508999973535538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.031757891178131104,
      "epoch": 8.04,
      "learning_rate": 4.7359911794501526e-05,
      "loss": 0.0294,
      "step": 8464,
      "task_loss": 0.007956236600875854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1145702600479126,
      "epoch": 8.04,
      "learning_rate": 4.7355143938409785e-05,
      "loss": 0.1207,
      "step": 8465,
      "task_loss": 0.175959050655365
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1442631483078003,
      "epoch": 8.04,
      "learning_rate": 4.735037202139849e-05,
      "loss": 0.1416,
      "step": 8466,
      "task_loss": 0.1179547980427742
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05049190670251846,
      "epoch": 8.04,
      "learning_rate": 4.734559604433447e-05,
      "loss": 0.0476,
      "step": 8467,
      "task_loss": 0.0210769884288311
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021424848586320877,
      "epoch": 8.04,
      "learning_rate": 4.734081600808531e-05,
      "loss": 0.0274,
      "step": 8468,
      "task_loss": 0.08106733858585358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027790479362010956,
      "epoch": 8.04,
      "learning_rate": 4.733603191351933e-05,
      "loss": 0.0255,
      "step": 8469,
      "task_loss": 0.004662582650780678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021773390471935272,
      "epoch": 8.04,
      "learning_rate": 4.733124376150558e-05,
      "loss": 0.027,
      "step": 8470,
      "task_loss": 0.07355040311813354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0780288502573967,
      "epoch": 8.04,
      "learning_rate": 4.7326451552913856e-05,
      "loss": 0.0713,
      "step": 8471,
      "task_loss": 0.011100053787231445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07223173975944519,
      "epoch": 8.05,
      "learning_rate": 4.7321655288614674e-05,
      "loss": 0.0675,
      "step": 8472,
      "task_loss": 0.024789290502667427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020628787577152252,
      "epoch": 8.05,
      "learning_rate": 4.7316854969479314e-05,
      "loss": 0.0269,
      "step": 8473,
      "task_loss": 0.08376266062259674
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08520027995109558,
      "epoch": 8.05,
      "learning_rate": 4.7312050596379764e-05,
      "loss": 0.0812,
      "step": 8474,
      "task_loss": 0.0456637404859066
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021483324468135834,
      "epoch": 8.05,
      "learning_rate": 4.730724217018877e-05,
      "loss": 0.0244,
      "step": 8475,
      "task_loss": 0.050539180636405945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.28899598121643066,
      "epoch": 8.05,
      "learning_rate": 4.7302429691779806e-05,
      "loss": 0.2756,
      "step": 8476,
      "task_loss": 0.154689222574234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016797779127955437,
      "epoch": 8.05,
      "learning_rate": 4.729761316202708e-05,
      "loss": 0.0163,
      "step": 8477,
      "task_loss": 0.011920711025595665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02568567357957363,
      "epoch": 8.05,
      "learning_rate": 4.729279258180553e-05,
      "loss": 0.0242,
      "step": 8478,
      "task_loss": 0.011267339810729027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04046610742807388,
      "epoch": 8.05,
      "learning_rate": 4.7287967951990855e-05,
      "loss": 0.043,
      "step": 8479,
      "task_loss": 0.06550759077072144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.061091069132089615,
      "epoch": 8.05,
      "learning_rate": 4.7283139273459445e-05,
      "loss": 0.0739,
      "step": 8480,
      "task_loss": 0.1896328330039978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09908688813447952,
      "epoch": 8.05,
      "learning_rate": 4.727830654708848e-05,
      "loss": 0.0966,
      "step": 8481,
      "task_loss": 0.07403028011322021
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03582286834716797,
      "epoch": 8.06,
      "learning_rate": 4.727346977375584e-05,
      "loss": 0.0419,
      "step": 8482,
      "task_loss": 0.09690375626087189
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015889644622802734,
      "epoch": 8.06,
      "learning_rate": 4.7268628954340136e-05,
      "loss": 0.016,
      "step": 8483,
      "task_loss": 0.017025936394929886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013576588593423367,
      "epoch": 8.06,
      "learning_rate": 4.726378408972074e-05,
      "loss": 0.0387,
      "step": 8484,
      "task_loss": 0.26450613141059875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10775915533304214,
      "epoch": 8.06,
      "learning_rate": 4.725893518077774e-05,
      "loss": 0.1018,
      "step": 8485,
      "task_loss": 0.047988370060920715
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04508550837635994,
      "epoch": 8.06,
      "learning_rate": 4.725408222839197e-05,
      "loss": 0.0458,
      "step": 8486,
      "task_loss": 0.05229977145791054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01839440129697323,
      "epoch": 8.06,
      "learning_rate": 4.724922523344498e-05,
      "loss": 0.0171,
      "step": 8487,
      "task_loss": 0.0054893046617507935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.17300604283809662,
      "epoch": 8.06,
      "learning_rate": 4.724436419681907e-05,
      "loss": 0.165,
      "step": 8488,
      "task_loss": 0.09258658438920975
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.2892667055130005,
      "epoch": 8.06,
      "learning_rate": 4.723949911939728e-05,
      "loss": 0.2899,
      "step": 8489,
      "task_loss": 0.2951303720474243
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10540339350700378,
      "epoch": 8.06,
      "learning_rate": 4.723463000206337e-05,
      "loss": 0.102,
      "step": 8490,
      "task_loss": 0.07126626372337341
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04866549000144005,
      "epoch": 8.06,
      "learning_rate": 4.722975684570183e-05,
      "loss": 0.0448,
      "step": 8491,
      "task_loss": 0.009562673047184944
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.043563053011894226,
      "epoch": 8.06,
      "learning_rate": 4.7224879651197905e-05,
      "loss": 0.0401,
      "step": 8492,
      "task_loss": 0.00909213162958622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12174250930547714,
      "epoch": 8.07,
      "learning_rate": 4.721999841943755e-05,
      "loss": 0.1173,
      "step": 8493,
      "task_loss": 0.07682164013385773
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019868122413754463,
      "epoch": 8.07,
      "learning_rate": 4.721511315130747e-05,
      "loss": 0.029,
      "step": 8494,
      "task_loss": 0.11105664074420929
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04917510598897934,
      "epoch": 8.07,
      "learning_rate": 4.7210223847695104e-05,
      "loss": 0.0623,
      "step": 8495,
      "task_loss": 0.1808261275291443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.2701743543148041,
      "epoch": 8.07,
      "learning_rate": 4.72053305094886e-05,
      "loss": 0.2656,
      "step": 8496,
      "task_loss": 0.22409585118293762
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03515856713056564,
      "epoch": 8.07,
      "learning_rate": 4.720043313757687e-05,
      "loss": 0.0465,
      "step": 8497,
      "task_loss": 0.14875555038452148
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.20637600123882294,
      "epoch": 8.07,
      "learning_rate": 4.719553173284955e-05,
      "loss": 0.2016,
      "step": 8498,
      "task_loss": 0.15902316570281982
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10677587240934372,
      "epoch": 8.07,
      "learning_rate": 4.719062629619699e-05,
      "loss": 0.1103,
      "step": 8499,
      "task_loss": 0.14183999598026276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11615284532308578,
      "epoch": 8.07,
      "learning_rate": 4.71857168285103e-05,
      "loss": 0.1115,
      "step": 8500,
      "task_loss": 0.07010817527770996
    },
    {
      "epoch": 8.07,
      "eval_accuracy": 0.8864678899082569,
      "eval_loss": 0.4948354661464691,
      "eval_runtime": 18.1296,
      "eval_samples_per_second": 48.098,
      "eval_steps_per_second": 6.012,
      "step": 8500
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10728442668914795,
      "epoch": 8.07,
      "learning_rate": 4.718080333068129e-05,
      "loss": 0.1144,
      "step": 8501,
      "task_loss": 0.17879854142665863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.15180592238903046,
      "epoch": 8.07,
      "learning_rate": 4.717588580360253e-05,
      "loss": 0.1669,
      "step": 8502,
      "task_loss": 0.30242693424224854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03224493935704231,
      "epoch": 8.08,
      "learning_rate": 4.717096424816731e-05,
      "loss": 0.0302,
      "step": 8503,
      "task_loss": 0.012275317683815956
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.18254992365837097,
      "epoch": 8.08,
      "learning_rate": 4.716603866526967e-05,
      "loss": 0.1913,
      "step": 8504,
      "task_loss": 0.2695552706718445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01585567556321621,
      "epoch": 8.08,
      "learning_rate": 4.7161109055804356e-05,
      "loss": 0.0241,
      "step": 8505,
      "task_loss": 0.09879221767187119
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017822248861193657,
      "epoch": 8.08,
      "learning_rate": 4.7156175420666844e-05,
      "loss": 0.0263,
      "step": 8506,
      "task_loss": 0.10284404456615448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05827868729829788,
      "epoch": 8.08,
      "learning_rate": 4.715123776075336e-05,
      "loss": 0.0578,
      "step": 8507,
      "task_loss": 0.05304684489965439
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08329082280397415,
      "epoch": 8.08,
      "learning_rate": 4.714629607696086e-05,
      "loss": 0.0791,
      "step": 8508,
      "task_loss": 0.041491199284791946
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01390088815242052,
      "epoch": 8.08,
      "learning_rate": 4.714135037018702e-05,
      "loss": 0.0129,
      "step": 8509,
      "task_loss": 0.004111597314476967
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07340320199728012,
      "epoch": 8.08,
      "learning_rate": 4.713640064133025e-05,
      "loss": 0.0721,
      "step": 8510,
      "task_loss": 0.06043552607297897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023112181574106216,
      "epoch": 8.08,
      "learning_rate": 4.7131446891289694e-05,
      "loss": 0.0217,
      "step": 8511,
      "task_loss": 0.008507607504725456
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.165248841047287,
      "epoch": 8.08,
      "learning_rate": 4.712648912096522e-05,
      "loss": 0.1713,
      "step": 8512,
      "task_loss": 0.2260192334651947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.18116983771324158,
      "epoch": 8.08,
      "learning_rate": 4.712152733125744e-05,
      "loss": 0.1778,
      "step": 8513,
      "task_loss": 0.14782485365867615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03632102161645889,
      "epoch": 8.09,
      "learning_rate": 4.711656152306768e-05,
      "loss": 0.0373,
      "step": 8514,
      "task_loss": 0.04582914710044861
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01714969053864479,
      "epoch": 8.09,
      "learning_rate": 4.711159169729801e-05,
      "loss": 0.0161,
      "step": 8515,
      "task_loss": 0.006688836961984634
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13296061754226685,
      "epoch": 8.09,
      "learning_rate": 4.710661785485121e-05,
      "loss": 0.1435,
      "step": 8516,
      "task_loss": 0.23862020671367645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017725428566336632,
      "epoch": 8.09,
      "learning_rate": 4.710163999663081e-05,
      "loss": 0.0163,
      "step": 8517,
      "task_loss": 0.003909563645720482
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.2052067518234253,
      "epoch": 8.09,
      "learning_rate": 4.709665812354107e-05,
      "loss": 0.1971,
      "step": 8518,
      "task_loss": 0.12428037822246552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022109489887952805,
      "epoch": 8.09,
      "learning_rate": 4.709167223648695e-05,
      "loss": 0.0271,
      "step": 8519,
      "task_loss": 0.07243168354034424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01610475778579712,
      "epoch": 8.09,
      "learning_rate": 4.7086682336374187e-05,
      "loss": 0.0271,
      "step": 8520,
      "task_loss": 0.1261139065027237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1497061550617218,
      "epoch": 8.09,
      "learning_rate": 4.70816884241092e-05,
      "loss": 0.1447,
      "step": 8521,
      "task_loss": 0.09966839849948883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05508185923099518,
      "epoch": 8.09,
      "learning_rate": 4.7076690500599164e-05,
      "loss": 0.0549,
      "step": 8522,
      "task_loss": 0.05369473248720169
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022330323234200478,
      "epoch": 8.09,
      "learning_rate": 4.707168856675198e-05,
      "loss": 0.0208,
      "step": 8523,
      "task_loss": 0.007413491606712341
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03679078072309494,
      "epoch": 8.09,
      "learning_rate": 4.7066682623476265e-05,
      "loss": 0.0414,
      "step": 8524,
      "task_loss": 0.08316943049430847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016818320378661156,
      "epoch": 8.1,
      "learning_rate": 4.706167267168138e-05,
      "loss": 0.0156,
      "step": 8525,
      "task_loss": 0.004215966910123825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026049071922898293,
      "epoch": 8.1,
      "learning_rate": 4.70566587122774e-05,
      "loss": 0.0302,
      "step": 8526,
      "task_loss": 0.06748848408460617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12002403289079666,
      "epoch": 8.1,
      "learning_rate": 4.7051640746175147e-05,
      "loss": 0.1124,
      "step": 8527,
      "task_loss": 0.04354723170399666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1447361707687378,
      "epoch": 8.1,
      "learning_rate": 4.7046618774286146e-05,
      "loss": 0.1348,
      "step": 8528,
      "task_loss": 0.04555728659033775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05529174208641052,
      "epoch": 8.1,
      "learning_rate": 4.7041592797522664e-05,
      "loss": 0.0689,
      "step": 8529,
      "task_loss": 0.19117003679275513
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.032082926481962204,
      "epoch": 8.1,
      "learning_rate": 4.7036562816797705e-05,
      "loss": 0.0319,
      "step": 8530,
      "task_loss": 0.030074482783675194
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04032071679830551,
      "epoch": 8.1,
      "learning_rate": 4.7031528833024976e-05,
      "loss": 0.0484,
      "step": 8531,
      "task_loss": 0.12074785679578781
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.19668981432914734,
      "epoch": 8.1,
      "learning_rate": 4.702649084711892e-05,
      "loss": 0.1871,
      "step": 8532,
      "task_loss": 0.10096022486686707
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09332388639450073,
      "epoch": 8.1,
      "learning_rate": 4.7021448859994735e-05,
      "loss": 0.0903,
      "step": 8533,
      "task_loss": 0.06286977231502533
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07967086136341095,
      "epoch": 8.1,
      "learning_rate": 4.70164028725683e-05,
      "loss": 0.089,
      "step": 8534,
      "task_loss": 0.17311781644821167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04477657005190849,
      "epoch": 8.11,
      "learning_rate": 4.7011352885756255e-05,
      "loss": 0.0428,
      "step": 8535,
      "task_loss": 0.025373056530952454
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09086555242538452,
      "epoch": 8.11,
      "learning_rate": 4.7006298900475954e-05,
      "loss": 0.0939,
      "step": 8536,
      "task_loss": 0.12131273001432419
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021383512765169144,
      "epoch": 8.11,
      "learning_rate": 4.7001240917645465e-05,
      "loss": 0.0204,
      "step": 8537,
      "task_loss": 0.011899461969733238
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11136257648468018,
      "epoch": 8.11,
      "learning_rate": 4.699617893818361e-05,
      "loss": 0.1068,
      "step": 8538,
      "task_loss": 0.06574281305074692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.052705250680446625,
      "epoch": 8.11,
      "learning_rate": 4.699111296300992e-05,
      "loss": 0.0526,
      "step": 8539,
      "task_loss": 0.05185743421316147
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04489322006702423,
      "epoch": 8.11,
      "learning_rate": 4.6986042993044645e-05,
      "loss": 0.0547,
      "step": 8540,
      "task_loss": 0.14266639947891235
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11219049990177155,
      "epoch": 8.11,
      "learning_rate": 4.698096902920877e-05,
      "loss": 0.1172,
      "step": 8541,
      "task_loss": 0.1626092791557312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04092048108577728,
      "epoch": 8.11,
      "learning_rate": 4.6975891072424015e-05,
      "loss": 0.041,
      "step": 8542,
      "task_loss": 0.04180833697319031
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023674312978982925,
      "epoch": 8.11,
      "learning_rate": 4.697080912361281e-05,
      "loss": 0.0218,
      "step": 8543,
      "task_loss": 0.004959875717759132
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014916968531906605,
      "epoch": 8.11,
      "learning_rate": 4.696572318369831e-05,
      "loss": 0.0276,
      "step": 8544,
      "task_loss": 0.14138737320899963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02858028933405876,
      "epoch": 8.11,
      "learning_rate": 4.696063325360441e-05,
      "loss": 0.0384,
      "step": 8545,
      "task_loss": 0.1269652247428894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1576182246208191,
      "epoch": 8.12,
      "learning_rate": 4.6955539334255716e-05,
      "loss": 0.1567,
      "step": 8546,
      "task_loss": 0.14879369735717773
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.060055363923311234,
      "epoch": 8.12,
      "learning_rate": 4.6950441426577565e-05,
      "loss": 0.0705,
      "step": 8547,
      "task_loss": 0.16406583786010742
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07873189449310303,
      "epoch": 8.12,
      "learning_rate": 4.694533953149601e-05,
      "loss": 0.0885,
      "step": 8548,
      "task_loss": 0.17685380578041077
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015233471989631653,
      "epoch": 8.12,
      "learning_rate": 4.694023364993784e-05,
      "loss": 0.0165,
      "step": 8549,
      "task_loss": 0.027818040922284126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01854429766535759,
      "epoch": 8.12,
      "learning_rate": 4.693512378283056e-05,
      "loss": 0.0253,
      "step": 8550,
      "task_loss": 0.08647675812244415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05369843915104866,
      "epoch": 8.12,
      "learning_rate": 4.693000993110241e-05,
      "loss": 0.049,
      "step": 8551,
      "task_loss": 0.006441434845328331
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0389072448015213,
      "epoch": 8.12,
      "learning_rate": 4.692489209568234e-05,
      "loss": 0.0412,
      "step": 8552,
      "task_loss": 0.06141179800033569
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015452216379344463,
      "epoch": 8.12,
      "learning_rate": 4.691977027750002e-05,
      "loss": 0.0143,
      "step": 8553,
      "task_loss": 0.004333069548010826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01049085147678852,
      "epoch": 8.12,
      "learning_rate": 4.691464447748587e-05,
      "loss": 0.0097,
      "step": 8554,
      "task_loss": 0.003054805099964142
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08345725387334824,
      "epoch": 8.12,
      "learning_rate": 4.690951469657101e-05,
      "loss": 0.0808,
      "step": 8555,
      "task_loss": 0.05647343769669533
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013090891763567924,
      "epoch": 8.13,
      "learning_rate": 4.690438093568728e-05,
      "loss": 0.0123,
      "step": 8556,
      "task_loss": 0.005195150151848793
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03142661973834038,
      "epoch": 8.13,
      "learning_rate": 4.689924319576727e-05,
      "loss": 0.0294,
      "step": 8557,
      "task_loss": 0.010683547705411911
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12756206095218658,
      "epoch": 8.13,
      "learning_rate": 4.689410147774426e-05,
      "loss": 0.1162,
      "step": 8558,
      "task_loss": 0.014250561594963074
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04624927043914795,
      "epoch": 8.13,
      "learning_rate": 4.6888955782552274e-05,
      "loss": 0.0421,
      "step": 8559,
      "task_loss": 0.00475945882499218
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025834165513515472,
      "epoch": 8.13,
      "learning_rate": 4.688380611112605e-05,
      "loss": 0.0244,
      "step": 8560,
      "task_loss": 0.011195512488484383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02918969839811325,
      "epoch": 8.13,
      "learning_rate": 4.687865246440106e-05,
      "loss": 0.0374,
      "step": 8561,
      "task_loss": 0.11153009533882141
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012696018442511559,
      "epoch": 8.13,
      "learning_rate": 4.687349484331347e-05,
      "loss": 0.0118,
      "step": 8562,
      "task_loss": 0.0034159980714321136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01882791891694069,
      "epoch": 8.13,
      "learning_rate": 4.6868333248800204e-05,
      "loss": 0.0241,
      "step": 8563,
      "task_loss": 0.07115612179040909
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1464093029499054,
      "epoch": 8.13,
      "learning_rate": 4.686316768179889e-05,
      "loss": 0.1556,
      "step": 8564,
      "task_loss": 0.23828445374965668
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0876997858285904,
      "epoch": 8.13,
      "learning_rate": 4.685799814324786e-05,
      "loss": 0.091,
      "step": 8565,
      "task_loss": 0.12048880755901337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.2233988344669342,
      "epoch": 8.13,
      "learning_rate": 4.685282463408621e-05,
      "loss": 0.2209,
      "step": 8566,
      "task_loss": 0.198834627866745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02312629297375679,
      "epoch": 8.14,
      "learning_rate": 4.6847647155253716e-05,
      "loss": 0.0287,
      "step": 8567,
      "task_loss": 0.0787849947810173
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024075020104646683,
      "epoch": 8.14,
      "learning_rate": 4.684246570769089e-05,
      "loss": 0.0264,
      "step": 8568,
      "task_loss": 0.047648604959249496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1928960680961609,
      "epoch": 8.14,
      "learning_rate": 4.683728029233898e-05,
      "loss": 0.1874,
      "step": 8569,
      "task_loss": 0.13750508427619934
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09075068682432175,
      "epoch": 8.14,
      "learning_rate": 4.683209091013994e-05,
      "loss": 0.0963,
      "step": 8570,
      "task_loss": 0.1463644802570343
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04680275171995163,
      "epoch": 8.14,
      "learning_rate": 4.682689756203643e-05,
      "loss": 0.0503,
      "step": 8571,
      "task_loss": 0.08158842474222183
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02860000915825367,
      "epoch": 8.14,
      "learning_rate": 4.682170024897187e-05,
      "loss": 0.0327,
      "step": 8572,
      "task_loss": 0.06912478804588318
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08807064592838287,
      "epoch": 8.14,
      "learning_rate": 4.681649897189036e-05,
      "loss": 0.0828,
      "step": 8573,
      "task_loss": 0.03544972091913223
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03363920748233795,
      "epoch": 8.14,
      "learning_rate": 4.681129373173674e-05,
      "loss": 0.0422,
      "step": 8574,
      "task_loss": 0.11956753581762314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04612544924020767,
      "epoch": 8.14,
      "learning_rate": 4.6806084529456574e-05,
      "loss": 0.044,
      "step": 8575,
      "task_loss": 0.025233760476112366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09824679046869278,
      "epoch": 8.14,
      "learning_rate": 4.6800871365996135e-05,
      "loss": 0.0946,
      "step": 8576,
      "task_loss": 0.06206507235765457
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08123184740543365,
      "epoch": 8.15,
      "learning_rate": 4.679565424230241e-05,
      "loss": 0.0874,
      "step": 8577,
      "task_loss": 0.143352210521698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03590530529618263,
      "epoch": 8.15,
      "learning_rate": 4.679043315932313e-05,
      "loss": 0.04,
      "step": 8578,
      "task_loss": 0.07692985236644745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026768725365400314,
      "epoch": 8.15,
      "learning_rate": 4.6785208118006715e-05,
      "loss": 0.0329,
      "step": 8579,
      "task_loss": 0.08837291598320007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0306844525039196,
      "epoch": 8.15,
      "learning_rate": 4.677997911930234e-05,
      "loss": 0.0395,
      "step": 8580,
      "task_loss": 0.11881330609321594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03350158408284187,
      "epoch": 8.15,
      "learning_rate": 4.6774746164159854e-05,
      "loss": 0.0419,
      "step": 8581,
      "task_loss": 0.11749888956546783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.16534090042114258,
      "epoch": 8.15,
      "learning_rate": 4.676950925352986e-05,
      "loss": 0.1613,
      "step": 8582,
      "task_loss": 0.1252642273902893
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01491392683237791,
      "epoch": 8.15,
      "learning_rate": 4.676426838836367e-05,
      "loss": 0.0203,
      "step": 8583,
      "task_loss": 0.06893758475780487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04224181920289993,
      "epoch": 8.15,
      "learning_rate": 4.675902356961331e-05,
      "loss": 0.0482,
      "step": 8584,
      "task_loss": 0.10220709443092346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024668492376804352,
      "epoch": 8.15,
      "learning_rate": 4.675377479823153e-05,
      "loss": 0.033,
      "step": 8585,
      "task_loss": 0.10758410394191742
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04424036294221878,
      "epoch": 8.15,
      "learning_rate": 4.6748522075171784e-05,
      "loss": 0.0558,
      "step": 8586,
      "task_loss": 0.15995724499225616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05094079673290253,
      "epoch": 8.15,
      "learning_rate": 4.674326540138826e-05,
      "loss": 0.0489,
      "step": 8587,
      "task_loss": 0.030478911474347115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.14906425774097443,
      "epoch": 8.16,
      "learning_rate": 4.673800477783587e-05,
      "loss": 0.1457,
      "step": 8588,
      "task_loss": 0.11561848223209381
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.051018789410591125,
      "epoch": 8.16,
      "learning_rate": 4.6732740205470206e-05,
      "loss": 0.0608,
      "step": 8589,
      "task_loss": 0.14838165044784546
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03625892847776413,
      "epoch": 8.16,
      "learning_rate": 4.672747168524762e-05,
      "loss": 0.0397,
      "step": 8590,
      "task_loss": 0.07113449275493622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025677144527435303,
      "epoch": 8.16,
      "learning_rate": 4.672219921812517e-05,
      "loss": 0.0392,
      "step": 8591,
      "task_loss": 0.16079393029212952
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08316448330879211,
      "epoch": 8.16,
      "learning_rate": 4.671692280506061e-05,
      "loss": 0.0914,
      "step": 8592,
      "task_loss": 0.16506913304328918
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09451714903116226,
      "epoch": 8.16,
      "learning_rate": 4.671164244701243e-05,
      "loss": 0.0976,
      "step": 8593,
      "task_loss": 0.12491125613451004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.24787138402462006,
      "epoch": 8.16,
      "learning_rate": 4.670635814493984e-05,
      "loss": 0.2439,
      "step": 8594,
      "task_loss": 0.20846538245677948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0107728848233819,
      "epoch": 8.16,
      "learning_rate": 4.6701069899802755e-05,
      "loss": 0.01,
      "step": 8595,
      "task_loss": 0.0030124839395284653
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12116492539644241,
      "epoch": 8.16,
      "learning_rate": 4.669577771256181e-05,
      "loss": 0.1183,
      "step": 8596,
      "task_loss": 0.09291059523820877
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.034399181604385376,
      "epoch": 8.16,
      "learning_rate": 4.6690481584178354e-05,
      "loss": 0.0317,
      "step": 8597,
      "task_loss": 0.007727684453129768
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0222720205783844,
      "epoch": 8.17,
      "learning_rate": 4.6685181515614454e-05,
      "loss": 0.0206,
      "step": 8598,
      "task_loss": 0.0058763641864061356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07261856645345688,
      "epoch": 8.17,
      "learning_rate": 4.6679877507832895e-05,
      "loss": 0.0695,
      "step": 8599,
      "task_loss": 0.041326478123664856
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07069084793329239,
      "epoch": 8.17,
      "learning_rate": 4.6674569561797174e-05,
      "loss": 0.0644,
      "step": 8600,
      "task_loss": 0.0074882470071315765
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.037158627063035965,
      "epoch": 8.17,
      "learning_rate": 4.666925767847151e-05,
      "loss": 0.0424,
      "step": 8601,
      "task_loss": 0.08922519534826279
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03928046673536301,
      "epoch": 8.17,
      "learning_rate": 4.6663941858820825e-05,
      "loss": 0.051,
      "step": 8602,
      "task_loss": 0.15673944354057312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.040694400668144226,
      "epoch": 8.17,
      "learning_rate": 4.665862210381077e-05,
      "loss": 0.0384,
      "step": 8603,
      "task_loss": 0.01785385236144066
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06909865885972977,
      "epoch": 8.17,
      "learning_rate": 4.66532984144077e-05,
      "loss": 0.0671,
      "step": 8604,
      "task_loss": 0.04940890148282051
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027597632259130478,
      "epoch": 8.17,
      "learning_rate": 4.6647970791578685e-05,
      "loss": 0.0295,
      "step": 8605,
      "task_loss": 0.0467480830848217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015055635944008827,
      "epoch": 8.17,
      "learning_rate": 4.664263923629153e-05,
      "loss": 0.014,
      "step": 8606,
      "task_loss": 0.0045957863330841064
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11752206087112427,
      "epoch": 8.17,
      "learning_rate": 4.663730374951472e-05,
      "loss": 0.1067,
      "step": 8607,
      "task_loss": 0.009500864893198013
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07718470692634583,
      "epoch": 8.17,
      "learning_rate": 4.663196433221747e-05,
      "loss": 0.0789,
      "step": 8608,
      "task_loss": 0.09422293305397034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027676379308104515,
      "epoch": 8.18,
      "learning_rate": 4.6626620985369724e-05,
      "loss": 0.0254,
      "step": 8609,
      "task_loss": 0.004925228655338287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05825607106089592,
      "epoch": 8.18,
      "learning_rate": 4.662127370994212e-05,
      "loss": 0.0602,
      "step": 8610,
      "task_loss": 0.07727469503879547
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029302431270480156,
      "epoch": 8.18,
      "learning_rate": 4.6615922506906016e-05,
      "loss": 0.0288,
      "step": 8611,
      "task_loss": 0.024653693661093712
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1494738757610321,
      "epoch": 8.18,
      "learning_rate": 4.661056737723349e-05,
      "loss": 0.1421,
      "step": 8612,
      "task_loss": 0.07620874047279358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0703209936618805,
      "epoch": 8.18,
      "learning_rate": 4.660520832189732e-05,
      "loss": 0.0738,
      "step": 8613,
      "task_loss": 0.10534647107124329
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02900974452495575,
      "epoch": 8.18,
      "learning_rate": 4.6599845341871005e-05,
      "loss": 0.0268,
      "step": 8614,
      "task_loss": 0.007358167320489883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013132540509104729,
      "epoch": 8.18,
      "learning_rate": 4.6594478438128757e-05,
      "loss": 0.0122,
      "step": 8615,
      "task_loss": 0.003975642845034599
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05000152438879013,
      "epoch": 8.18,
      "learning_rate": 4.6589107611645497e-05,
      "loss": 0.0534,
      "step": 8616,
      "task_loss": 0.08441969007253647
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014517636969685555,
      "epoch": 8.18,
      "learning_rate": 4.658373286339688e-05,
      "loss": 0.0181,
      "step": 8617,
      "task_loss": 0.050333477556705475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11546745896339417,
      "epoch": 8.18,
      "learning_rate": 4.6578354194359227e-05,
      "loss": 0.1114,
      "step": 8618,
      "task_loss": 0.07436027377843857
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03974412381649017,
      "epoch": 8.19,
      "learning_rate": 4.657297160550961e-05,
      "loss": 0.0471,
      "step": 8619,
      "task_loss": 0.11349973827600479
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02339399978518486,
      "epoch": 8.19,
      "learning_rate": 4.656758509782582e-05,
      "loss": 0.0319,
      "step": 8620,
      "task_loss": 0.10848347842693329
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03902260586619377,
      "epoch": 8.19,
      "learning_rate": 4.6562194672286306e-05,
      "loss": 0.0458,
      "step": 8621,
      "task_loss": 0.10702133923768997
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029453573748469353,
      "epoch": 8.19,
      "learning_rate": 4.65568003298703e-05,
      "loss": 0.0416,
      "step": 8622,
      "task_loss": 0.15059900283813477
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0717625841498375,
      "epoch": 8.19,
      "learning_rate": 4.655140207155769e-05,
      "loss": 0.0746,
      "step": 8623,
      "task_loss": 0.09989339113235474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03731407970190048,
      "epoch": 8.19,
      "learning_rate": 4.65459998983291e-05,
      "loss": 0.0491,
      "step": 8624,
      "task_loss": 0.15505677461624146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0684528648853302,
      "epoch": 8.19,
      "learning_rate": 4.6540593811165866e-05,
      "loss": 0.0718,
      "step": 8625,
      "task_loss": 0.10180987417697906
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.18652784824371338,
      "epoch": 8.19,
      "learning_rate": 4.653518381105002e-05,
      "loss": 0.1783,
      "step": 8626,
      "task_loss": 0.10443468391895294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0480349101126194,
      "epoch": 8.19,
      "learning_rate": 4.6529769898964325e-05,
      "loss": 0.0462,
      "step": 8627,
      "task_loss": 0.02956531010568142
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09778586030006409,
      "epoch": 8.19,
      "learning_rate": 4.652435207589224e-05,
      "loss": 0.1007,
      "step": 8628,
      "task_loss": 0.12683331966400146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025784488767385483,
      "epoch": 8.19,
      "learning_rate": 4.651893034281793e-05,
      "loss": 0.0257,
      "step": 8629,
      "task_loss": 0.024955110624432564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05094904452562332,
      "epoch": 8.2,
      "learning_rate": 4.6513504700726293e-05,
      "loss": 0.065,
      "step": 8630,
      "task_loss": 0.19132231175899506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10476318001747131,
      "epoch": 8.2,
      "learning_rate": 4.650807515060291e-05,
      "loss": 0.1067,
      "step": 8631,
      "task_loss": 0.12431478500366211
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.17144785821437836,
      "epoch": 8.2,
      "learning_rate": 4.650264169343411e-05,
      "loss": 0.1651,
      "step": 8632,
      "task_loss": 0.108132004737854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04371756315231323,
      "epoch": 8.2,
      "learning_rate": 4.6497204330206874e-05,
      "loss": 0.0494,
      "step": 8633,
      "task_loss": 0.10011433064937592
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024459153413772583,
      "epoch": 8.2,
      "learning_rate": 4.649176306190895e-05,
      "loss": 0.0227,
      "step": 8634,
      "task_loss": 0.0073354970663785934
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030715011060237885,
      "epoch": 8.2,
      "learning_rate": 4.648631788952874e-05,
      "loss": 0.0359,
      "step": 8635,
      "task_loss": 0.0821438580751419
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07763121277093887,
      "epoch": 8.2,
      "learning_rate": 4.6480868814055424e-05,
      "loss": 0.0733,
      "step": 8636,
      "task_loss": 0.0338205024600029
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029433991760015488,
      "epoch": 8.2,
      "learning_rate": 4.647541583647883e-05,
      "loss": 0.0364,
      "step": 8637,
      "task_loss": 0.09859539568424225
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.2271520048379898,
      "epoch": 8.2,
      "learning_rate": 4.646995895778952e-05,
      "loss": 0.2236,
      "step": 8638,
      "task_loss": 0.19156351685523987
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0869775339961052,
      "epoch": 8.2,
      "learning_rate": 4.646449817897876e-05,
      "loss": 0.0906,
      "step": 8639,
      "task_loss": 0.1231326162815094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024121176451444626,
      "epoch": 8.21,
      "learning_rate": 4.645903350103855e-05,
      "loss": 0.0275,
      "step": 8640,
      "task_loss": 0.058099135756492615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019324442371726036,
      "epoch": 8.21,
      "learning_rate": 4.6453564924961544e-05,
      "loss": 0.0184,
      "step": 8641,
      "task_loss": 0.010152887552976608
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01817592792212963,
      "epoch": 8.21,
      "learning_rate": 4.644809245174114e-05,
      "loss": 0.0266,
      "step": 8642,
      "task_loss": 0.10227106511592865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02699386328458786,
      "epoch": 8.21,
      "learning_rate": 4.6442616082371466e-05,
      "loss": 0.0387,
      "step": 8643,
      "task_loss": 0.14438271522521973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0430140346288681,
      "epoch": 8.21,
      "learning_rate": 4.64371358178473e-05,
      "loss": 0.0394,
      "step": 8644,
      "task_loss": 0.006518969312310219
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03487422317266464,
      "epoch": 8.21,
      "learning_rate": 4.6431651659164174e-05,
      "loss": 0.0456,
      "step": 8645,
      "task_loss": 0.14229948818683624
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027154915034770966,
      "epoch": 8.21,
      "learning_rate": 4.6426163607318305e-05,
      "loss": 0.0251,
      "step": 8646,
      "task_loss": 0.006713952869176865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03345032036304474,
      "epoch": 8.21,
      "learning_rate": 4.642067166330663e-05,
      "loss": 0.0407,
      "step": 8647,
      "task_loss": 0.10630976408720016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030117200687527657,
      "epoch": 8.21,
      "learning_rate": 4.6415175828126786e-05,
      "loss": 0.0276,
      "step": 8648,
      "task_loss": 0.004965195432305336
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08309546858072281,
      "epoch": 8.21,
      "learning_rate": 4.640967610277711e-05,
      "loss": 0.0893,
      "step": 8649,
      "task_loss": 0.14526864886283875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10631243884563446,
      "epoch": 8.21,
      "learning_rate": 4.640417248825667e-05,
      "loss": 0.102,
      "step": 8650,
      "task_loss": 0.0636465847492218
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04527665674686432,
      "epoch": 8.22,
      "learning_rate": 4.63986649855652e-05,
      "loss": 0.0411,
      "step": 8651,
      "task_loss": 0.003472359851002693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1377941519021988,
      "epoch": 8.22,
      "learning_rate": 4.639315359570319e-05,
      "loss": 0.1355,
      "step": 8652,
      "task_loss": 0.11462458968162537
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03246547654271126,
      "epoch": 8.22,
      "learning_rate": 4.6387638319671786e-05,
      "loss": 0.0305,
      "step": 8653,
      "task_loss": 0.013004310429096222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028419774025678635,
      "epoch": 8.22,
      "learning_rate": 4.6382119158472895e-05,
      "loss": 0.0263,
      "step": 8654,
      "task_loss": 0.006932957097887993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017502637580037117,
      "epoch": 8.22,
      "learning_rate": 4.637659611310907e-05,
      "loss": 0.0174,
      "step": 8655,
      "task_loss": 0.01657920889556408
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13791508972644806,
      "epoch": 8.22,
      "learning_rate": 4.637106918458361e-05,
      "loss": 0.1373,
      "step": 8656,
      "task_loss": 0.1319553405046463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.055541716516017914,
      "epoch": 8.22,
      "learning_rate": 4.636553837390051e-05,
      "loss": 0.0648,
      "step": 8657,
      "task_loss": 0.14844538271427155
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10020337253808975,
      "epoch": 8.22,
      "learning_rate": 4.636000368206447e-05,
      "loss": 0.0989,
      "step": 8658,
      "task_loss": 0.08745051920413971
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09979698061943054,
      "epoch": 8.22,
      "learning_rate": 4.6354465110080885e-05,
      "loss": 0.0986,
      "step": 8659,
      "task_loss": 0.08759049326181412
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12549327313899994,
      "epoch": 8.22,
      "learning_rate": 4.6348922658955874e-05,
      "loss": 0.1235,
      "step": 8660,
      "task_loss": 0.10560569912195206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11727667599916458,
      "epoch": 8.23,
      "learning_rate": 4.634337632969624e-05,
      "loss": 0.1162,
      "step": 8661,
      "task_loss": 0.10662281513214111
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04554348438978195,
      "epoch": 8.23,
      "learning_rate": 4.6337826123309505e-05,
      "loss": 0.0446,
      "step": 8662,
      "task_loss": 0.035856109112501144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.052027251571416855,
      "epoch": 8.23,
      "learning_rate": 4.6332272040803895e-05,
      "loss": 0.0485,
      "step": 8663,
      "task_loss": 0.016855746507644653
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0639544203877449,
      "epoch": 8.23,
      "learning_rate": 4.632671408318833e-05,
      "loss": 0.0608,
      "step": 8664,
      "task_loss": 0.03275006264448166
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07394473999738693,
      "epoch": 8.23,
      "learning_rate": 4.6321152251472435e-05,
      "loss": 0.081,
      "step": 8665,
      "task_loss": 0.14465902745723724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07188697159290314,
      "epoch": 8.23,
      "learning_rate": 4.6315586546666556e-05,
      "loss": 0.0734,
      "step": 8666,
      "task_loss": 0.08725833147764206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023782558739185333,
      "epoch": 8.23,
      "learning_rate": 4.631001696978172e-05,
      "loss": 0.0278,
      "step": 8667,
      "task_loss": 0.06402748823165894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.032316289842128754,
      "epoch": 8.23,
      "learning_rate": 4.630444352182968e-05,
      "loss": 0.0384,
      "step": 8668,
      "task_loss": 0.09317083656787872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030993333086371422,
      "epoch": 8.23,
      "learning_rate": 4.6298866203822865e-05,
      "loss": 0.0293,
      "step": 8669,
      "task_loss": 0.01442483440041542
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11551865190267563,
      "epoch": 8.23,
      "learning_rate": 4.629328501677442e-05,
      "loss": 0.1106,
      "step": 8670,
      "task_loss": 0.06652466952800751
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10263466089963913,
      "epoch": 8.23,
      "learning_rate": 4.6287699961698214e-05,
      "loss": 0.0992,
      "step": 8671,
      "task_loss": 0.06787226349115372
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04704195261001587,
      "epoch": 8.24,
      "learning_rate": 4.6282111039608784e-05,
      "loss": 0.0512,
      "step": 8672,
      "task_loss": 0.08842041343450546
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11540669202804565,
      "epoch": 8.24,
      "learning_rate": 4.6276518251521384e-05,
      "loss": 0.1093,
      "step": 8673,
      "task_loss": 0.05394323915243149
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019262634217739105,
      "epoch": 8.24,
      "learning_rate": 4.6270921598451974e-05,
      "loss": 0.0268,
      "step": 8674,
      "task_loss": 0.09500335156917572
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06018088012933731,
      "epoch": 8.24,
      "learning_rate": 4.6265321081417223e-05,
      "loss": 0.0591,
      "step": 8675,
      "task_loss": 0.049118801951408386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13014157116413116,
      "epoch": 8.24,
      "learning_rate": 4.625971670143447e-05,
      "loss": 0.1193,
      "step": 8676,
      "task_loss": 0.02204965241253376
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024654846638441086,
      "epoch": 8.24,
      "learning_rate": 4.625410845952181e-05,
      "loss": 0.023,
      "step": 8677,
      "task_loss": 0.007745366543531418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02617044746875763,
      "epoch": 8.24,
      "learning_rate": 4.6248496356697966e-05,
      "loss": 0.0267,
      "step": 8678,
      "task_loss": 0.03177504613995552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022578658536076546,
      "epoch": 8.24,
      "learning_rate": 4.6242880393982436e-05,
      "loss": 0.0278,
      "step": 8679,
      "task_loss": 0.07478535175323486
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06433162093162537,
      "epoch": 8.24,
      "learning_rate": 4.623726057239537e-05,
      "loss": 0.0829,
      "step": 8680,
      "task_loss": 0.24952758848667145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028324808925390244,
      "epoch": 8.24,
      "learning_rate": 4.623163689295764e-05,
      "loss": 0.0271,
      "step": 8681,
      "task_loss": 0.01628146879374981
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03707346320152283,
      "epoch": 8.25,
      "learning_rate": 4.6226009356690825e-05,
      "loss": 0.0343,
      "step": 8682,
      "task_loss": 0.009730465710163116
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0707152709364891,
      "epoch": 8.25,
      "learning_rate": 4.622037796461718e-05,
      "loss": 0.0825,
      "step": 8683,
      "task_loss": 0.18847694993019104
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015443956479430199,
      "epoch": 8.25,
      "learning_rate": 4.621474271775968e-05,
      "loss": 0.0152,
      "step": 8684,
      "task_loss": 0.013239886611700058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0768551155924797,
      "epoch": 8.25,
      "learning_rate": 4.620910361714199e-05,
      "loss": 0.0735,
      "step": 8685,
      "task_loss": 0.043199822306632996
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04744374752044678,
      "epoch": 8.25,
      "learning_rate": 4.620346066378849e-05,
      "loss": 0.0446,
      "step": 8686,
      "task_loss": 0.018627936020493507
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03644431009888649,
      "epoch": 8.25,
      "learning_rate": 4.619781385872424e-05,
      "loss": 0.0407,
      "step": 8687,
      "task_loss": 0.07931828498840332
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028153453022241592,
      "epoch": 8.25,
      "learning_rate": 4.6192163202975013e-05,
      "loss": 0.0388,
      "step": 8688,
      "task_loss": 0.1342998743057251
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02188189886510372,
      "epoch": 8.25,
      "learning_rate": 4.618650869756728e-05,
      "loss": 0.0249,
      "step": 8689,
      "task_loss": 0.0519079715013504
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1566876918077469,
      "epoch": 8.25,
      "learning_rate": 4.6180850343528205e-05,
      "loss": 0.1496,
      "step": 8690,
      "task_loss": 0.08567800372838974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04210164025425911,
      "epoch": 8.25,
      "learning_rate": 4.617518814188565e-05,
      "loss": 0.0436,
      "step": 8691,
      "task_loss": 0.0568234808743
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019453078508377075,
      "epoch": 8.25,
      "learning_rate": 4.6169522093668196e-05,
      "loss": 0.0268,
      "step": 8692,
      "task_loss": 0.09267456084489822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028510289266705513,
      "epoch": 8.26,
      "learning_rate": 4.61638521999051e-05,
      "loss": 0.0422,
      "step": 8693,
      "task_loss": 0.16586299240589142
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09608705341815948,
      "epoch": 8.26,
      "learning_rate": 4.6158178461626323e-05,
      "loss": 0.0922,
      "step": 8694,
      "task_loss": 0.057251427322626114
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.035421546548604965,
      "epoch": 8.26,
      "learning_rate": 4.615250087986254e-05,
      "loss": 0.0374,
      "step": 8695,
      "task_loss": 0.05525943636894226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03646400570869446,
      "epoch": 8.26,
      "learning_rate": 4.6146819455645086e-05,
      "loss": 0.0426,
      "step": 8696,
      "task_loss": 0.09792999178171158
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08208326995372772,
      "epoch": 8.26,
      "learning_rate": 4.614113419000604e-05,
      "loss": 0.0778,
      "step": 8697,
      "task_loss": 0.0388401597738266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01685364544391632,
      "epoch": 8.26,
      "learning_rate": 4.613544508397815e-05,
      "loss": 0.0169,
      "step": 8698,
      "task_loss": 0.017076315358281136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02594437077641487,
      "epoch": 8.26,
      "learning_rate": 4.6129752138594874e-05,
      "loss": 0.0407,
      "step": 8699,
      "task_loss": 0.17328821122646332
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02614816278219223,
      "epoch": 8.26,
      "learning_rate": 4.612405535489036e-05,
      "loss": 0.0368,
      "step": 8700,
      "task_loss": 0.1322673261165619
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.037083324044942856,
      "epoch": 8.26,
      "learning_rate": 4.611835473389945e-05,
      "loss": 0.034,
      "step": 8701,
      "task_loss": 0.005980234593153
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019921965897083282,
      "epoch": 8.26,
      "learning_rate": 4.61126502766577e-05,
      "loss": 0.0183,
      "step": 8702,
      "task_loss": 0.0037397872656583786
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07418224960565567,
      "epoch": 8.26,
      "learning_rate": 4.6106941984201344e-05,
      "loss": 0.0676,
      "step": 8703,
      "task_loss": 0.008437564596533775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0171127300709486,
      "epoch": 8.27,
      "learning_rate": 4.610122985756733e-05,
      "loss": 0.0281,
      "step": 8704,
      "task_loss": 0.12655523419380188
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.041877150535583496,
      "epoch": 8.27,
      "learning_rate": 4.609551389779328e-05,
      "loss": 0.0504,
      "step": 8705,
      "task_loss": 0.12753605842590332
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09651815891265869,
      "epoch": 8.27,
      "learning_rate": 4.6089794105917544e-05,
      "loss": 0.0924,
      "step": 8706,
      "task_loss": 0.05498852580785751
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026990918442606926,
      "epoch": 8.27,
      "learning_rate": 4.6084070482979135e-05,
      "loss": 0.0247,
      "step": 8707,
      "task_loss": 0.00403929129242897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01692625693976879,
      "epoch": 8.27,
      "learning_rate": 4.607834303001778e-05,
      "loss": 0.0239,
      "step": 8708,
      "task_loss": 0.08626563102006912
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02325316146016121,
      "epoch": 8.27,
      "learning_rate": 4.60726117480739e-05,
      "loss": 0.0215,
      "step": 8709,
      "task_loss": 0.005591306835412979
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012971121817827225,
      "epoch": 8.27,
      "learning_rate": 4.6066876638188604e-05,
      "loss": 0.0121,
      "step": 8710,
      "task_loss": 0.0038487426936626434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10507965087890625,
      "epoch": 8.27,
      "learning_rate": 4.606113770140371e-05,
      "loss": 0.1015,
      "step": 8711,
      "task_loss": 0.06916863471269608
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04460505023598671,
      "epoch": 8.27,
      "learning_rate": 4.605539493876173e-05,
      "loss": 0.0626,
      "step": 8712,
      "task_loss": 0.224493145942688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13093966245651245,
      "epoch": 8.27,
      "learning_rate": 4.604964835130585e-05,
      "loss": 0.1438,
      "step": 8713,
      "task_loss": 0.2592054605484009
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.054761841893196106,
      "epoch": 8.28,
      "learning_rate": 4.6043897940079964e-05,
      "loss": 0.0584,
      "step": 8714,
      "task_loss": 0.0908980518579483
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10678750276565552,
      "epoch": 8.28,
      "learning_rate": 4.603814370612867e-05,
      "loss": 0.1026,
      "step": 8715,
      "task_loss": 0.06508254259824753
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04329557716846466,
      "epoch": 8.28,
      "learning_rate": 4.603238565049726e-05,
      "loss": 0.047,
      "step": 8716,
      "task_loss": 0.08073478937149048
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01392638310790062,
      "epoch": 8.28,
      "learning_rate": 4.60266237742317e-05,
      "loss": 0.013,
      "step": 8717,
      "task_loss": 0.00464806891977787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1740546077489853,
      "epoch": 8.28,
      "learning_rate": 4.602085807837866e-05,
      "loss": 0.1735,
      "step": 8718,
      "task_loss": 0.1681438386440277
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1236797422170639,
      "epoch": 8.28,
      "learning_rate": 4.601508856398552e-05,
      "loss": 0.1305,
      "step": 8719,
      "task_loss": 0.19167181849479675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.043940044939517975,
      "epoch": 8.28,
      "learning_rate": 4.6009315232100324e-05,
      "loss": 0.0636,
      "step": 8720,
      "task_loss": 0.2409691959619522
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09241293370723724,
      "epoch": 8.28,
      "learning_rate": 4.600353808377184e-05,
      "loss": 0.0951,
      "step": 8721,
      "task_loss": 0.119747593998909
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020489653572440147,
      "epoch": 8.28,
      "learning_rate": 4.599775712004951e-05,
      "loss": 0.0191,
      "step": 8722,
      "task_loss": 0.0063454341143369675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09082825481891632,
      "epoch": 8.28,
      "learning_rate": 4.599197234198347e-05,
      "loss": 0.0912,
      "step": 8723,
      "task_loss": 0.09437683969736099
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029831204563379288,
      "epoch": 8.28,
      "learning_rate": 4.5986183750624555e-05,
      "loss": 0.0506,
      "step": 8724,
      "task_loss": 0.23745323717594147
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06204962357878685,
      "epoch": 8.29,
      "learning_rate": 4.5980391347024296e-05,
      "loss": 0.062,
      "step": 8725,
      "task_loss": 0.061779238283634186
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03659776970744133,
      "epoch": 8.29,
      "learning_rate": 4.59745951322349e-05,
      "loss": 0.0351,
      "step": 8726,
      "task_loss": 0.02147502824664116
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024865522980690002,
      "epoch": 8.29,
      "learning_rate": 4.596879510730929e-05,
      "loss": 0.0345,
      "step": 8727,
      "task_loss": 0.1214059591293335
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04416849464178085,
      "epoch": 8.29,
      "learning_rate": 4.596299127330106e-05,
      "loss": 0.0599,
      "step": 8728,
      "task_loss": 0.20144742727279663
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06300773471593857,
      "epoch": 8.29,
      "learning_rate": 4.59571836312645e-05,
      "loss": 0.0675,
      "step": 8729,
      "task_loss": 0.10772567242383957
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03947564959526062,
      "epoch": 8.29,
      "learning_rate": 4.595137218225461e-05,
      "loss": 0.0452,
      "step": 8730,
      "task_loss": 0.09653180092573166
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021150944754481316,
      "epoch": 8.29,
      "learning_rate": 4.594555692732706e-05,
      "loss": 0.0268,
      "step": 8731,
      "task_loss": 0.07796701788902283
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.17825676500797272,
      "epoch": 8.29,
      "learning_rate": 4.593973786753821e-05,
      "loss": 0.1831,
      "step": 8732,
      "task_loss": 0.22676941752433777
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012715199962258339,
      "epoch": 8.29,
      "learning_rate": 4.593391500394514e-05,
      "loss": 0.0187,
      "step": 8733,
      "task_loss": 0.07283098250627518
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13977791368961334,
      "epoch": 8.29,
      "learning_rate": 4.5928088337605586e-05,
      "loss": 0.1398,
      "step": 8734,
      "task_loss": 0.1397220343351364
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.14722847938537598,
      "epoch": 8.3,
      "learning_rate": 4.5922257869578e-05,
      "loss": 0.1469,
      "step": 8735,
      "task_loss": 0.14371134340763092
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.15708275139331818,
      "epoch": 8.3,
      "learning_rate": 4.5916423600921496e-05,
      "loss": 0.1519,
      "step": 8736,
      "task_loss": 0.10510636121034622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.057722412049770355,
      "epoch": 8.3,
      "learning_rate": 4.591058553269593e-05,
      "loss": 0.0712,
      "step": 8737,
      "task_loss": 0.1927632987499237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04139891266822815,
      "epoch": 8.3,
      "learning_rate": 4.590474366596178e-05,
      "loss": 0.0431,
      "step": 8738,
      "task_loss": 0.05822301283478737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021710364148020744,
      "epoch": 8.3,
      "learning_rate": 4.589889800178026e-05,
      "loss": 0.0253,
      "step": 8739,
      "task_loss": 0.05777839943766594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022335700690746307,
      "epoch": 8.3,
      "learning_rate": 4.589304854121329e-05,
      "loss": 0.0211,
      "step": 8740,
      "task_loss": 0.009688341990113258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026100587099790573,
      "epoch": 8.3,
      "learning_rate": 4.588719528532342e-05,
      "loss": 0.0327,
      "step": 8741,
      "task_loss": 0.09173382818698883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12068310379981995,
      "epoch": 8.3,
      "learning_rate": 4.588133823517392e-05,
      "loss": 0.115,
      "step": 8742,
      "task_loss": 0.06337210536003113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025651680305600166,
      "epoch": 8.3,
      "learning_rate": 4.587547739182878e-05,
      "loss": 0.0238,
      "step": 8743,
      "task_loss": 0.006972752511501312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11778855323791504,
      "epoch": 8.3,
      "learning_rate": 4.586961275635263e-05,
      "loss": 0.1207,
      "step": 8744,
      "task_loss": 0.1473734825849533
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.14205443859100342,
      "epoch": 8.3,
      "learning_rate": 4.586374432981081e-05,
      "loss": 0.1341,
      "step": 8745,
      "task_loss": 0.06203455477952957
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018492717295885086,
      "epoch": 8.31,
      "learning_rate": 4.585787211326935e-05,
      "loss": 0.0172,
      "step": 8746,
      "task_loss": 0.0057936906814575195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10602115839719772,
      "epoch": 8.31,
      "learning_rate": 4.5851996107794975e-05,
      "loss": 0.1199,
      "step": 8747,
      "task_loss": 0.24453692138195038
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022601492702960968,
      "epoch": 8.31,
      "learning_rate": 4.584611631445508e-05,
      "loss": 0.0279,
      "step": 8748,
      "task_loss": 0.07518120110034943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05565394088625908,
      "epoch": 8.31,
      "learning_rate": 4.5840232734317754e-05,
      "loss": 0.0597,
      "step": 8749,
      "task_loss": 0.09581439942121506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.035259947180747986,
      "epoch": 8.31,
      "learning_rate": 4.583434536845179e-05,
      "loss": 0.0394,
      "step": 8750,
      "task_loss": 0.07706654071807861
    },
    {
      "epoch": 8.31,
      "eval_accuracy": 0.8956422018348624,
      "eval_loss": 0.4291454553604126,
      "eval_runtime": 17.9719,
      "eval_samples_per_second": 48.52,
      "eval_steps_per_second": 6.065,
      "step": 8750
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06929297000169754,
      "epoch": 8.31,
      "learning_rate": 4.5828454217926654e-05,
      "loss": 0.0863,
      "step": 8751,
      "task_loss": 0.23925699293613434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06375081837177277,
      "epoch": 8.31,
      "learning_rate": 4.5822559283812496e-05,
      "loss": 0.0626,
      "step": 8752,
      "task_loss": 0.05265332758426666
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11460275948047638,
      "epoch": 8.31,
      "learning_rate": 4.581666056718016e-05,
      "loss": 0.1133,
      "step": 8753,
      "task_loss": 0.10144174098968506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0539216548204422,
      "epoch": 8.31,
      "learning_rate": 4.5810758069101175e-05,
      "loss": 0.052,
      "step": 8754,
      "task_loss": 0.034788183867931366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030059363692998886,
      "epoch": 8.31,
      "learning_rate": 4.580485179064777e-05,
      "loss": 0.0336,
      "step": 8755,
      "task_loss": 0.06580464541912079
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1027214527130127,
      "epoch": 8.32,
      "learning_rate": 4.579894173289284e-05,
      "loss": 0.1106,
      "step": 8756,
      "task_loss": 0.18194906413555145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08241204172372818,
      "epoch": 8.32,
      "learning_rate": 4.579302789690997e-05,
      "loss": 0.0806,
      "step": 8757,
      "task_loss": 0.0640886202454567
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016954604536294937,
      "epoch": 8.32,
      "learning_rate": 4.578711028377344e-05,
      "loss": 0.0219,
      "step": 8758,
      "task_loss": 0.06677691638469696
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02651621401309967,
      "epoch": 8.32,
      "learning_rate": 4.578118889455821e-05,
      "loss": 0.0253,
      "step": 8759,
      "task_loss": 0.014431167393922806
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03452229127287865,
      "epoch": 8.32,
      "learning_rate": 4.577526373033994e-05,
      "loss": 0.0328,
      "step": 8760,
      "task_loss": 0.017721977084875107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08960497379302979,
      "epoch": 8.32,
      "learning_rate": 4.576933479219496e-05,
      "loss": 0.0924,
      "step": 8761,
      "task_loss": 0.11729642748832703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07891800254583359,
      "epoch": 8.32,
      "learning_rate": 4.5763402081200294e-05,
      "loss": 0.0769,
      "step": 8762,
      "task_loss": 0.05859693884849548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02912452071905136,
      "epoch": 8.32,
      "learning_rate": 4.575746559843364e-05,
      "loss": 0.0273,
      "step": 8763,
      "task_loss": 0.010460572317242622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028819631785154343,
      "epoch": 8.32,
      "learning_rate": 4.5751525344973384e-05,
      "loss": 0.0329,
      "step": 8764,
      "task_loss": 0.06981261819601059
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05606947839260101,
      "epoch": 8.32,
      "learning_rate": 4.5745581321898615e-05,
      "loss": 0.0654,
      "step": 8765,
      "task_loss": 0.14961570501327515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019708652049303055,
      "epoch": 8.32,
      "learning_rate": 4.5739633530289085e-05,
      "loss": 0.0288,
      "step": 8766,
      "task_loss": 0.1107863336801529
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.043510615825653076,
      "epoch": 8.33,
      "learning_rate": 4.573368197122524e-05,
      "loss": 0.0518,
      "step": 8767,
      "task_loss": 0.12616392970085144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04988428205251694,
      "epoch": 8.33,
      "learning_rate": 4.572772664578821e-05,
      "loss": 0.0522,
      "step": 8768,
      "task_loss": 0.07312482595443726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.047488268464803696,
      "epoch": 8.33,
      "learning_rate": 4.572176755505981e-05,
      "loss": 0.0509,
      "step": 8769,
      "task_loss": 0.08157598972320557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05487481504678726,
      "epoch": 8.33,
      "learning_rate": 4.571580470012254e-05,
      "loss": 0.0548,
      "step": 8770,
      "task_loss": 0.05453791469335556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02108554169535637,
      "epoch": 8.33,
      "learning_rate": 4.5709838082059574e-05,
      "loss": 0.0193,
      "step": 8771,
      "task_loss": 0.002928614616394043
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09459959715604782,
      "epoch": 8.33,
      "learning_rate": 4.570386770195478e-05,
      "loss": 0.0858,
      "step": 8772,
      "task_loss": 0.006600510329008102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028554365038871765,
      "epoch": 8.33,
      "learning_rate": 4.569789356089271e-05,
      "loss": 0.0308,
      "step": 8773,
      "task_loss": 0.05108209699392319
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.051416561007499695,
      "epoch": 8.33,
      "learning_rate": 4.569191565995859e-05,
      "loss": 0.0613,
      "step": 8774,
      "task_loss": 0.15025511384010315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020000001415610313,
      "epoch": 8.33,
      "learning_rate": 4.568593400023834e-05,
      "loss": 0.0228,
      "step": 8775,
      "task_loss": 0.047657888382673264
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03664660453796387,
      "epoch": 8.33,
      "learning_rate": 4.567994858281855e-05,
      "loss": 0.0338,
      "step": 8776,
      "task_loss": 0.008631901815533638
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08228196948766708,
      "epoch": 8.34,
      "learning_rate": 4.56739594087865e-05,
      "loss": 0.0783,
      "step": 8777,
      "task_loss": 0.042297106236219406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.035452522337436676,
      "epoch": 8.34,
      "learning_rate": 4.566796647923017e-05,
      "loss": 0.0449,
      "step": 8778,
      "task_loss": 0.13000640273094177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011653019115328789,
      "epoch": 8.34,
      "learning_rate": 4.566196979523818e-05,
      "loss": 0.0197,
      "step": 8779,
      "task_loss": 0.09172951430082321
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08766936510801315,
      "epoch": 8.34,
      "learning_rate": 4.5655969357899874e-05,
      "loss": 0.0842,
      "step": 8780,
      "task_loss": 0.05248169228434563
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08307437598705292,
      "epoch": 8.34,
      "learning_rate": 4.564996516830525e-05,
      "loss": 0.0812,
      "step": 8781,
      "task_loss": 0.06468552350997925
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01311748381704092,
      "epoch": 8.34,
      "learning_rate": 4.564395722754501e-05,
      "loss": 0.0121,
      "step": 8782,
      "task_loss": 0.0029160063713788986
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022205941379070282,
      "epoch": 8.34,
      "learning_rate": 4.56379455367105e-05,
      "loss": 0.0262,
      "step": 8783,
      "task_loss": 0.061669085174798965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029047932475805283,
      "epoch": 8.34,
      "learning_rate": 4.563193009689381e-05,
      "loss": 0.0371,
      "step": 8784,
      "task_loss": 0.10945844650268555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019876539707183838,
      "epoch": 8.34,
      "learning_rate": 4.562591090918764e-05,
      "loss": 0.0275,
      "step": 8785,
      "task_loss": 0.09562841057777405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01689576543867588,
      "epoch": 8.34,
      "learning_rate": 4.561988797468542e-05,
      "loss": 0.0234,
      "step": 8786,
      "task_loss": 0.08180932700634003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023341480642557144,
      "epoch": 8.34,
      "learning_rate": 4.561386129448125e-05,
      "loss": 0.0305,
      "step": 8787,
      "task_loss": 0.09533084183931351
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015885740518569946,
      "epoch": 8.35,
      "learning_rate": 4.5607830869669885e-05,
      "loss": 0.0212,
      "step": 8788,
      "task_loss": 0.06913591921329498
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026201194152235985,
      "epoch": 8.35,
      "learning_rate": 4.560179670134681e-05,
      "loss": 0.025,
      "step": 8789,
      "task_loss": 0.013938097283244133
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08920443058013916,
      "epoch": 8.35,
      "learning_rate": 4.559575879060813e-05,
      "loss": 0.0968,
      "step": 8790,
      "task_loss": 0.16520185768604279
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1296762377023697,
      "epoch": 8.35,
      "learning_rate": 4.5589717138550685e-05,
      "loss": 0.1242,
      "step": 8791,
      "task_loss": 0.07465289533138275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016865486279129982,
      "epoch": 8.35,
      "learning_rate": 4.5583671746271964e-05,
      "loss": 0.0156,
      "step": 8792,
      "task_loss": 0.00429266132414341
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030185125768184662,
      "epoch": 8.35,
      "learning_rate": 4.557762261487013e-05,
      "loss": 0.0461,
      "step": 8793,
      "task_loss": 0.18932506442070007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.039688460528850555,
      "epoch": 8.35,
      "learning_rate": 4.557156974544404e-05,
      "loss": 0.037,
      "step": 8794,
      "task_loss": 0.013088133186101913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06738491356372833,
      "epoch": 8.35,
      "learning_rate": 4.5565513139093244e-05,
      "loss": 0.0756,
      "step": 8795,
      "task_loss": 0.1495150625705719
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05773467570543289,
      "epoch": 8.35,
      "learning_rate": 4.5559452796917936e-05,
      "loss": 0.0613,
      "step": 8796,
      "task_loss": 0.09331423789262772
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026063233613967896,
      "epoch": 8.35,
      "learning_rate": 4.555338872001901e-05,
      "loss": 0.0311,
      "step": 8797,
      "task_loss": 0.07620520889759064
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018153179436922073,
      "epoch": 8.36,
      "learning_rate": 4.554732090949805e-05,
      "loss": 0.0175,
      "step": 8798,
      "task_loss": 0.011946845799684525
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03151557967066765,
      "epoch": 8.36,
      "learning_rate": 4.5541249366457276e-05,
      "loss": 0.0288,
      "step": 8799,
      "task_loss": 0.0040736570954322815
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01914232410490513,
      "epoch": 8.36,
      "learning_rate": 4.5535174091999636e-05,
      "loss": 0.0221,
      "step": 8800,
      "task_loss": 0.04857531934976578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023833172395825386,
      "epoch": 8.36,
      "learning_rate": 4.552909508722871e-05,
      "loss": 0.0246,
      "step": 8801,
      "task_loss": 0.03120681270956993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021301060914993286,
      "epoch": 8.36,
      "learning_rate": 4.55230123532488e-05,
      "loss": 0.0197,
      "step": 8802,
      "task_loss": 0.005311897024512291
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.034260865300893784,
      "epoch": 8.36,
      "learning_rate": 4.551692589116486e-05,
      "loss": 0.0544,
      "step": 8803,
      "task_loss": 0.23611676692962646
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.036078743636608124,
      "epoch": 8.36,
      "learning_rate": 4.551083570208252e-05,
      "loss": 0.0371,
      "step": 8804,
      "task_loss": 0.0467095673084259
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.14268112182617188,
      "epoch": 8.36,
      "learning_rate": 4.550474178710809e-05,
      "loss": 0.1415,
      "step": 8805,
      "task_loss": 0.1312038004398346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014835448935627937,
      "epoch": 8.36,
      "learning_rate": 4.549864414734856e-05,
      "loss": 0.0139,
      "step": 8806,
      "task_loss": 0.005272580310702324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06314484775066376,
      "epoch": 8.36,
      "learning_rate": 4.54925427839116e-05,
      "loss": 0.0702,
      "step": 8807,
      "task_loss": 0.13396084308624268
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09572551399469376,
      "epoch": 8.36,
      "learning_rate": 4.548643769790556e-05,
      "loss": 0.0963,
      "step": 8808,
      "task_loss": 0.10133033990859985
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017699236050248146,
      "epoch": 8.37,
      "learning_rate": 4.548032889043944e-05,
      "loss": 0.0374,
      "step": 8809,
      "task_loss": 0.21495309472084045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05119137093424797,
      "epoch": 8.37,
      "learning_rate": 4.547421636262294e-05,
      "loss": 0.05,
      "step": 8810,
      "task_loss": 0.039462827146053314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024145277217030525,
      "epoch": 8.37,
      "learning_rate": 4.546810011556644e-05,
      "loss": 0.0334,
      "step": 8811,
      "task_loss": 0.11649482697248459
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10837046056985855,
      "epoch": 8.37,
      "learning_rate": 4.546198015038097e-05,
      "loss": 0.1182,
      "step": 8812,
      "task_loss": 0.20654115080833435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10691244900226593,
      "epoch": 8.37,
      "learning_rate": 4.545585646817826e-05,
      "loss": 0.1056,
      "step": 8813,
      "task_loss": 0.09399950504302979
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10578563064336777,
      "epoch": 8.37,
      "learning_rate": 4.544972907007071e-05,
      "loss": 0.1143,
      "step": 8814,
      "task_loss": 0.1912904679775238
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07613591104745865,
      "epoch": 8.37,
      "learning_rate": 4.544359795717139e-05,
      "loss": 0.0696,
      "step": 8815,
      "task_loss": 0.010670226067304611
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03094104677438736,
      "epoch": 8.37,
      "learning_rate": 4.543746313059404e-05,
      "loss": 0.0365,
      "step": 8816,
      "task_loss": 0.08633188903331757
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0631537064909935,
      "epoch": 8.37,
      "learning_rate": 4.5431324591453094e-05,
      "loss": 0.0812,
      "step": 8817,
      "task_loss": 0.2431773990392685
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.034764595329761505,
      "epoch": 8.37,
      "learning_rate": 4.5425182340863626e-05,
      "loss": 0.0319,
      "step": 8818,
      "task_loss": 0.006121266633272171
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.035581640899181366,
      "epoch": 8.38,
      "learning_rate": 4.541903637994142e-05,
      "loss": 0.033,
      "step": 8819,
      "task_loss": 0.009517015889286995
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08049920201301575,
      "epoch": 8.38,
      "learning_rate": 4.541288670980291e-05,
      "loss": 0.0904,
      "step": 8820,
      "task_loss": 0.17998726665973663
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06685604900121689,
      "epoch": 8.38,
      "learning_rate": 4.540673333156523e-05,
      "loss": 0.0659,
      "step": 8821,
      "task_loss": 0.057141952216625214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0435507632791996,
      "epoch": 8.38,
      "learning_rate": 4.540057624634616e-05,
      "loss": 0.0404,
      "step": 8822,
      "task_loss": 0.01237713173031807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03415733203291893,
      "epoch": 8.38,
      "learning_rate": 4.5394415455264164e-05,
      "loss": 0.0395,
      "step": 8823,
      "task_loss": 0.0872575044631958
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.060022905468940735,
      "epoch": 8.38,
      "learning_rate": 4.538825095943838e-05,
      "loss": 0.0712,
      "step": 8824,
      "task_loss": 0.17156316339969635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04664577543735504,
      "epoch": 8.38,
      "learning_rate": 4.538208275998861e-05,
      "loss": 0.05,
      "step": 8825,
      "task_loss": 0.08014590293169022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08388902246952057,
      "epoch": 8.38,
      "learning_rate": 4.537591085803535e-05,
      "loss": 0.0931,
      "step": 8826,
      "task_loss": 0.17568424344062805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02333504892885685,
      "epoch": 8.38,
      "learning_rate": 4.5369735254699754e-05,
      "loss": 0.0281,
      "step": 8827,
      "task_loss": 0.07060873508453369
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07975588738918304,
      "epoch": 8.38,
      "learning_rate": 4.536355595110365e-05,
      "loss": 0.0783,
      "step": 8828,
      "task_loss": 0.06561661511659622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05120707303285599,
      "epoch": 8.38,
      "learning_rate": 4.5357372948369534e-05,
      "loss": 0.0468,
      "step": 8829,
      "task_loss": 0.007422303780913353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08561455458402634,
      "epoch": 8.39,
      "learning_rate": 4.535118624762057e-05,
      "loss": 0.0831,
      "step": 8830,
      "task_loss": 0.06040613353252411
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02236170694231987,
      "epoch": 8.39,
      "learning_rate": 4.534499584998062e-05,
      "loss": 0.0205,
      "step": 8831,
      "task_loss": 0.00412105955183506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07289911806583405,
      "epoch": 8.39,
      "learning_rate": 4.533880175657419e-05,
      "loss": 0.0812,
      "step": 8832,
      "task_loss": 0.15584085881710052
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.103610560297966,
      "epoch": 8.39,
      "learning_rate": 4.533260396852646e-05,
      "loss": 0.1056,
      "step": 8833,
      "task_loss": 0.12309184670448303
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1255352646112442,
      "epoch": 8.39,
      "learning_rate": 4.532640248696331e-05,
      "loss": 0.1261,
      "step": 8834,
      "task_loss": 0.1312541514635086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028518887236714363,
      "epoch": 8.39,
      "learning_rate": 4.532019731301125e-05,
      "loss": 0.028,
      "step": 8835,
      "task_loss": 0.02283020317554474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04137152433395386,
      "epoch": 8.39,
      "learning_rate": 4.531398844779749e-05,
      "loss": 0.0637,
      "step": 8836,
      "task_loss": 0.26472553610801697
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13369058072566986,
      "epoch": 8.39,
      "learning_rate": 4.530777589244989e-05,
      "loss": 0.142,
      "step": 8837,
      "task_loss": 0.21714474260807037
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07951021194458008,
      "epoch": 8.39,
      "learning_rate": 4.5301559648096995e-05,
      "loss": 0.0786,
      "step": 8838,
      "task_loss": 0.0705207884311676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08249984681606293,
      "epoch": 8.39,
      "learning_rate": 4.529533971586802e-05,
      "loss": 0.0865,
      "step": 8839,
      "task_loss": 0.12243272364139557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06132117658853531,
      "epoch": 8.4,
      "learning_rate": 4.5289116096892834e-05,
      "loss": 0.0566,
      "step": 8840,
      "task_loss": 0.014352064579725266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.035899195820093155,
      "epoch": 8.4,
      "learning_rate": 4.5282888792302e-05,
      "loss": 0.0432,
      "step": 8841,
      "task_loss": 0.10917837917804718
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03924690932035446,
      "epoch": 8.4,
      "learning_rate": 4.527665780322674e-05,
      "loss": 0.042,
      "step": 8842,
      "task_loss": 0.06675869226455688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08691535890102386,
      "epoch": 8.4,
      "learning_rate": 4.527042313079893e-05,
      "loss": 0.0855,
      "step": 8843,
      "task_loss": 0.07236799597740173
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03491726517677307,
      "epoch": 8.4,
      "learning_rate": 4.526418477615114e-05,
      "loss": 0.0329,
      "step": 8844,
      "task_loss": 0.014603780582547188
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026119505986571312,
      "epoch": 8.4,
      "learning_rate": 4.525794274041658e-05,
      "loss": 0.0403,
      "step": 8845,
      "task_loss": 0.16816593706607819
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025338448584079742,
      "epoch": 8.4,
      "learning_rate": 4.5251697024729165e-05,
      "loss": 0.0233,
      "step": 8846,
      "task_loss": 0.0048566292971372604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.038359981030225754,
      "epoch": 8.4,
      "learning_rate": 4.524544763022346e-05,
      "loss": 0.0368,
      "step": 8847,
      "task_loss": 0.022422099485993385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.20507895946502686,
      "epoch": 8.4,
      "learning_rate": 4.523919455803468e-05,
      "loss": 0.1902,
      "step": 8848,
      "task_loss": 0.05593420937657356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02473161742091179,
      "epoch": 8.4,
      "learning_rate": 4.5232937809298734e-05,
      "loss": 0.0301,
      "step": 8849,
      "task_loss": 0.07794321328401566
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01353430561721325,
      "epoch": 8.4,
      "learning_rate": 4.5226677385152206e-05,
      "loss": 0.0133,
      "step": 8850,
      "task_loss": 0.011057652533054352
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029346950352191925,
      "epoch": 8.41,
      "learning_rate": 4.522041328673231e-05,
      "loss": 0.0364,
      "step": 8851,
      "task_loss": 0.09976962208747864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09129992872476578,
      "epoch": 8.41,
      "learning_rate": 4.521414551517695e-05,
      "loss": 0.0922,
      "step": 8852,
      "task_loss": 0.10062223672866821
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0642009973526001,
      "epoch": 8.41,
      "learning_rate": 4.520787407162471e-05,
      "loss": 0.0722,
      "step": 8853,
      "task_loss": 0.14382028579711914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02546757459640503,
      "epoch": 8.41,
      "learning_rate": 4.520159895721483e-05,
      "loss": 0.0233,
      "step": 8854,
      "task_loss": 0.0041826870292425156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1266893744468689,
      "epoch": 8.41,
      "learning_rate": 4.51953201730872e-05,
      "loss": 0.1269,
      "step": 8855,
      "task_loss": 0.12863604724407196
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04428763687610626,
      "epoch": 8.41,
      "learning_rate": 4.51890377203824e-05,
      "loss": 0.0412,
      "step": 8856,
      "task_loss": 0.013243492692708969
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008449184708297253,
      "epoch": 8.41,
      "learning_rate": 4.518275160024167e-05,
      "loss": 0.0082,
      "step": 8857,
      "task_loss": 0.006423516198992729
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08089472353458405,
      "epoch": 8.41,
      "learning_rate": 4.5176461813806904e-05,
      "loss": 0.0796,
      "step": 8858,
      "task_loss": 0.0684363842010498
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03204871341586113,
      "epoch": 8.41,
      "learning_rate": 4.5170168362220686e-05,
      "loss": 0.0306,
      "step": 8859,
      "task_loss": 0.017244910821318626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04458446055650711,
      "epoch": 8.41,
      "learning_rate": 4.516387124662624e-05,
      "loss": 0.048,
      "step": 8860,
      "task_loss": 0.07903842628002167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01822434365749359,
      "epoch": 8.42,
      "learning_rate": 4.5157570468167464e-05,
      "loss": 0.0169,
      "step": 8861,
      "task_loss": 0.005464507266879082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03550713509321213,
      "epoch": 8.42,
      "learning_rate": 4.5151266027988946e-05,
      "loss": 0.0333,
      "step": 8862,
      "task_loss": 0.013436004519462585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028216678649187088,
      "epoch": 8.42,
      "learning_rate": 4.51449579272359e-05,
      "loss": 0.0333,
      "step": 8863,
      "task_loss": 0.07930795848369598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01838553324341774,
      "epoch": 8.42,
      "learning_rate": 4.5138646167054224e-05,
      "loss": 0.017,
      "step": 8864,
      "task_loss": 0.004559867084026337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08109049499034882,
      "epoch": 8.42,
      "learning_rate": 4.513233074859049e-05,
      "loss": 0.0859,
      "step": 8865,
      "task_loss": 0.12965066730976105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016361292451620102,
      "epoch": 8.42,
      "learning_rate": 4.512601167299191e-05,
      "loss": 0.0212,
      "step": 8866,
      "task_loss": 0.06487929821014404
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011701793409883976,
      "epoch": 8.42,
      "learning_rate": 4.511968894140639e-05,
      "loss": 0.0264,
      "step": 8867,
      "task_loss": 0.15838071703910828
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06448012590408325,
      "epoch": 8.42,
      "learning_rate": 4.511336255498247e-05,
      "loss": 0.0657,
      "step": 8868,
      "task_loss": 0.07717397809028625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028120938688516617,
      "epoch": 8.42,
      "learning_rate": 4.510703251486937e-05,
      "loss": 0.026,
      "step": 8869,
      "task_loss": 0.006957884877920151
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10039125382900238,
      "epoch": 8.42,
      "learning_rate": 4.5100698822216984e-05,
      "loss": 0.1005,
      "step": 8870,
      "task_loss": 0.1014624685049057
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019519591704010963,
      "epoch": 8.42,
      "learning_rate": 4.509436147817585e-05,
      "loss": 0.018,
      "step": 8871,
      "task_loss": 0.0039330217987298965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.098009392619133,
      "epoch": 8.43,
      "learning_rate": 4.5088020483897184e-05,
      "loss": 0.102,
      "step": 8872,
      "task_loss": 0.13752353191375732
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09184578061103821,
      "epoch": 8.43,
      "learning_rate": 4.508167584053285e-05,
      "loss": 0.0878,
      "step": 8873,
      "task_loss": 0.05179011821746826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03356043994426727,
      "epoch": 8.43,
      "learning_rate": 4.507532754923537e-05,
      "loss": 0.0388,
      "step": 8874,
      "task_loss": 0.0859462320804596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08062370121479034,
      "epoch": 8.43,
      "learning_rate": 4.506897561115797e-05,
      "loss": 0.0896,
      "step": 8875,
      "task_loss": 0.17005644738674164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0661439374089241,
      "epoch": 8.43,
      "learning_rate": 4.506262002745449e-05,
      "loss": 0.0845,
      "step": 8876,
      "task_loss": 0.25010010600090027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04141310602426529,
      "epoch": 8.43,
      "learning_rate": 4.505626079927947e-05,
      "loss": 0.0399,
      "step": 8877,
      "task_loss": 0.025958728045225143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05498037487268448,
      "epoch": 8.43,
      "learning_rate": 4.504989792778808e-05,
      "loss": 0.0538,
      "step": 8878,
      "task_loss": 0.04316407069563866
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05957046523690224,
      "epoch": 8.43,
      "learning_rate": 4.504353141413616e-05,
      "loss": 0.0686,
      "step": 8879,
      "task_loss": 0.14945709705352783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03215145319700241,
      "epoch": 8.43,
      "learning_rate": 4.5037161259480246e-05,
      "loss": 0.0435,
      "step": 8880,
      "task_loss": 0.14541016519069672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03889644145965576,
      "epoch": 8.43,
      "learning_rate": 4.5030787464977476e-05,
      "loss": 0.036,
      "step": 8881,
      "task_loss": 0.009727858006954193
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028448857367038727,
      "epoch": 8.43,
      "learning_rate": 4.50244100317857e-05,
      "loss": 0.0262,
      "step": 8882,
      "task_loss": 0.005950525403022766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05188886076211929,
      "epoch": 8.44,
      "learning_rate": 4.5018028961063394e-05,
      "loss": 0.0695,
      "step": 8883,
      "task_loss": 0.22816702723503113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0323837585747242,
      "epoch": 8.44,
      "learning_rate": 4.501164425396973e-05,
      "loss": 0.0389,
      "step": 8884,
      "task_loss": 0.0973554477095604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06097978726029396,
      "epoch": 8.44,
      "learning_rate": 4.5005255911664507e-05,
      "loss": 0.0569,
      "step": 8885,
      "task_loss": 0.02063034474849701
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021500347182154655,
      "epoch": 8.44,
      "learning_rate": 4.49988639353082e-05,
      "loss": 0.0345,
      "step": 8886,
      "task_loss": 0.1510607749223709
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.053475480526685715,
      "epoch": 8.44,
      "learning_rate": 4.4992468326061944e-05,
      "loss": 0.0527,
      "step": 8887,
      "task_loss": 0.04602964594960213
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026177948340773582,
      "epoch": 8.44,
      "learning_rate": 4.498606908508754e-05,
      "loss": 0.0313,
      "step": 8888,
      "task_loss": 0.07776033878326416
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1273716241121292,
      "epoch": 8.44,
      "learning_rate": 4.4979666213547414e-05,
      "loss": 0.1253,
      "step": 8889,
      "task_loss": 0.10640902817249298
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028834780678153038,
      "epoch": 8.44,
      "learning_rate": 4.497325971260471e-05,
      "loss": 0.0271,
      "step": 8890,
      "task_loss": 0.011695507913827896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03636486828327179,
      "epoch": 8.44,
      "learning_rate": 4.496684958342319e-05,
      "loss": 0.0339,
      "step": 8891,
      "task_loss": 0.011735286563634872
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12257404625415802,
      "epoch": 8.44,
      "learning_rate": 4.4960435827167266e-05,
      "loss": 0.123,
      "step": 8892,
      "task_loss": 0.12667058408260345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027225444093346596,
      "epoch": 8.45,
      "learning_rate": 4.495401844500205e-05,
      "loss": 0.0297,
      "step": 8893,
      "task_loss": 0.051733896136283875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022315477952361107,
      "epoch": 8.45,
      "learning_rate": 4.494759743809329e-05,
      "loss": 0.025,
      "step": 8894,
      "task_loss": 0.04938486963510513
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05692111700773239,
      "epoch": 8.45,
      "learning_rate": 4.494117280760739e-05,
      "loss": 0.0631,
      "step": 8895,
      "task_loss": 0.11854865401983261
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017605263739824295,
      "epoch": 8.45,
      "learning_rate": 4.49347445547114e-05,
      "loss": 0.0404,
      "step": 8896,
      "task_loss": 0.24594470858573914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13257010281085968,
      "epoch": 8.45,
      "learning_rate": 4.4928312680573064e-05,
      "loss": 0.1377,
      "step": 8897,
      "task_loss": 0.18380433320999146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0691106989979744,
      "epoch": 8.45,
      "learning_rate": 4.492187718636075e-05,
      "loss": 0.0788,
      "step": 8898,
      "task_loss": 0.16619594395160675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03699222952127457,
      "epoch": 8.45,
      "learning_rate": 4.49154380732435e-05,
      "loss": 0.048,
      "step": 8899,
      "task_loss": 0.14703547954559326
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11394920945167542,
      "epoch": 8.45,
      "learning_rate": 4.490899534239101e-05,
      "loss": 0.1091,
      "step": 8900,
      "task_loss": 0.06529615074396133
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05552458390593529,
      "epoch": 8.45,
      "learning_rate": 4.490254899497364e-05,
      "loss": 0.0599,
      "step": 8901,
      "task_loss": 0.0996512770652771
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0950566753745079,
      "epoch": 8.45,
      "learning_rate": 4.4896099032162386e-05,
      "loss": 0.0994,
      "step": 8902,
      "task_loss": 0.13881003856658936
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029427465051412582,
      "epoch": 8.45,
      "learning_rate": 4.488964545512892e-05,
      "loss": 0.0426,
      "step": 8903,
      "task_loss": 0.16067925095558167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04379139095544815,
      "epoch": 8.46,
      "learning_rate": 4.488318826504557e-05,
      "loss": 0.0431,
      "step": 8904,
      "task_loss": 0.03657126426696777
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.053460512310266495,
      "epoch": 8.46,
      "learning_rate": 4.4876727463085324e-05,
      "loss": 0.0526,
      "step": 8905,
      "task_loss": 0.04512707516551018
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.036254942417144775,
      "epoch": 8.46,
      "learning_rate": 4.487026305042179e-05,
      "loss": 0.0371,
      "step": 8906,
      "task_loss": 0.045052558183670044
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015025094151496887,
      "epoch": 8.46,
      "learning_rate": 4.4863795028229286e-05,
      "loss": 0.0215,
      "step": 8907,
      "task_loss": 0.07951157540082932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01650727353990078,
      "epoch": 8.46,
      "learning_rate": 4.4857323397682746e-05,
      "loss": 0.0306,
      "step": 8908,
      "task_loss": 0.15717414021492004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06591572612524033,
      "epoch": 8.46,
      "learning_rate": 4.485084815995778e-05,
      "loss": 0.064,
      "step": 8909,
      "task_loss": 0.04637327417731285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09855660796165466,
      "epoch": 8.46,
      "learning_rate": 4.484436931623064e-05,
      "loss": 0.0949,
      "step": 8910,
      "task_loss": 0.06240401789546013
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027789302170276642,
      "epoch": 8.46,
      "learning_rate": 4.4837886867678245e-05,
      "loss": 0.0327,
      "step": 8911,
      "task_loss": 0.07687053829431534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07295016199350357,
      "epoch": 8.46,
      "learning_rate": 4.4831400815478164e-05,
      "loss": 0.0882,
      "step": 8912,
      "task_loss": 0.2257649004459381
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0313107855618,
      "epoch": 8.46,
      "learning_rate": 4.482491116080861e-05,
      "loss": 0.0417,
      "step": 8913,
      "task_loss": 0.1352921426296234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020768383517861366,
      "epoch": 8.47,
      "learning_rate": 4.4818417904848466e-05,
      "loss": 0.036,
      "step": 8914,
      "task_loss": 0.17357057332992554
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11295350641012192,
      "epoch": 8.47,
      "learning_rate": 4.481192104877726e-05,
      "loss": 0.1166,
      "step": 8915,
      "task_loss": 0.14907796680927277
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021822160109877586,
      "epoch": 8.47,
      "learning_rate": 4.480542059377519e-05,
      "loss": 0.0222,
      "step": 8916,
      "task_loss": 0.025608519092202187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02147882990539074,
      "epoch": 8.47,
      "learning_rate": 4.479891654102307e-05,
      "loss": 0.02,
      "step": 8917,
      "task_loss": 0.006244117394089699
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05350108444690704,
      "epoch": 8.47,
      "learning_rate": 4.4792408891702426e-05,
      "loss": 0.0564,
      "step": 8918,
      "task_loss": 0.08280838280916214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09106861054897308,
      "epoch": 8.47,
      "learning_rate": 4.4785897646995376e-05,
      "loss": 0.0881,
      "step": 8919,
      "task_loss": 0.06174633651971817
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028143450617790222,
      "epoch": 8.47,
      "learning_rate": 4.477938280808473e-05,
      "loss": 0.0261,
      "step": 8920,
      "task_loss": 0.007531605660915375
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028097132220864296,
      "epoch": 8.47,
      "learning_rate": 4.4772864376153936e-05,
      "loss": 0.0306,
      "step": 8921,
      "task_loss": 0.05265646427869797
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029321586713194847,
      "epoch": 8.47,
      "learning_rate": 4.4766342352387106e-05,
      "loss": 0.027,
      "step": 8922,
      "task_loss": 0.006089037284255028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06918927282094955,
      "epoch": 8.47,
      "learning_rate": 4.475981673796899e-05,
      "loss": 0.0718,
      "step": 8923,
      "task_loss": 0.09525958448648453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07651045173406601,
      "epoch": 8.47,
      "learning_rate": 4.475328753408499e-05,
      "loss": 0.0909,
      "step": 8924,
      "task_loss": 0.22058546543121338
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12579286098480225,
      "epoch": 8.48,
      "learning_rate": 4.474675474192119e-05,
      "loss": 0.1404,
      "step": 8925,
      "task_loss": 0.2715792953968048
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0216844342648983,
      "epoch": 8.48,
      "learning_rate": 4.4740218362664276e-05,
      "loss": 0.0204,
      "step": 8926,
      "task_loss": 0.008529262617230415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07038342207670212,
      "epoch": 8.48,
      "learning_rate": 4.473367839750165e-05,
      "loss": 0.0718,
      "step": 8927,
      "task_loss": 0.0848298892378807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.059562116861343384,
      "epoch": 8.48,
      "learning_rate": 4.4727134847621276e-05,
      "loss": 0.0718,
      "step": 8928,
      "task_loss": 0.18145687878131866
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07572756707668304,
      "epoch": 8.48,
      "learning_rate": 4.4720587714211863e-05,
      "loss": 0.0747,
      "step": 8929,
      "task_loss": 0.06551861763000488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06999503076076508,
      "epoch": 8.48,
      "learning_rate": 4.471403699846272e-05,
      "loss": 0.0659,
      "step": 8930,
      "task_loss": 0.0290969330817461
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028632836416363716,
      "epoch": 8.48,
      "learning_rate": 4.470748270156381e-05,
      "loss": 0.0337,
      "step": 8931,
      "task_loss": 0.07966382801532745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10344910621643066,
      "epoch": 8.48,
      "learning_rate": 4.4700924824705745e-05,
      "loss": 0.1,
      "step": 8932,
      "task_loss": 0.06940016150474548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09662455320358276,
      "epoch": 8.48,
      "learning_rate": 4.469436336907982e-05,
      "loss": 0.0968,
      "step": 8933,
      "task_loss": 0.09881898760795593
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08603687584400177,
      "epoch": 8.48,
      "learning_rate": 4.4687798335877936e-05,
      "loss": 0.0844,
      "step": 8934,
      "task_loss": 0.06977303326129913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05292690172791481,
      "epoch": 8.49,
      "learning_rate": 4.4681229726292664e-05,
      "loss": 0.0519,
      "step": 8935,
      "task_loss": 0.04311305657029152
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019087469205260277,
      "epoch": 8.49,
      "learning_rate": 4.4674657541517227e-05,
      "loss": 0.0191,
      "step": 8936,
      "task_loss": 0.018771233037114143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01726618967950344,
      "epoch": 8.49,
      "learning_rate": 4.466808178274549e-05,
      "loss": 0.0225,
      "step": 8937,
      "task_loss": 0.06974489986896515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07852751016616821,
      "epoch": 8.49,
      "learning_rate": 4.4661502451171975e-05,
      "loss": 0.0891,
      "step": 8938,
      "task_loss": 0.18472985923290253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015991097316145897,
      "epoch": 8.49,
      "learning_rate": 4.465491954799186e-05,
      "loss": 0.0148,
      "step": 8939,
      "task_loss": 0.004459971562027931
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07318315654993057,
      "epoch": 8.49,
      "learning_rate": 4.4648333074400936e-05,
      "loss": 0.0713,
      "step": 8940,
      "task_loss": 0.054741598665714264
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04243193194270134,
      "epoch": 8.49,
      "learning_rate": 4.464174303159569e-05,
      "loss": 0.0503,
      "step": 8941,
      "task_loss": 0.12083330750465393
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.18897034227848053,
      "epoch": 8.49,
      "learning_rate": 4.463514942077323e-05,
      "loss": 0.1905,
      "step": 8942,
      "task_loss": 0.20461271703243256
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09098904579877853,
      "epoch": 8.49,
      "learning_rate": 4.4628552243131304e-05,
      "loss": 0.0979,
      "step": 8943,
      "task_loss": 0.16002698242664337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021895185112953186,
      "epoch": 8.49,
      "learning_rate": 4.462195149986833e-05,
      "loss": 0.0259,
      "step": 8944,
      "task_loss": 0.061538372188806534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05251630023121834,
      "epoch": 8.49,
      "learning_rate": 4.4615347192183375e-05,
      "loss": 0.0511,
      "step": 8945,
      "task_loss": 0.038724854588508606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0883500799536705,
      "epoch": 8.5,
      "learning_rate": 4.4608739321276126e-05,
      "loss": 0.0855,
      "step": 8946,
      "task_loss": 0.060159243643283844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0961025282740593,
      "epoch": 8.5,
      "learning_rate": 4.4602127888346944e-05,
      "loss": 0.0918,
      "step": 8947,
      "task_loss": 0.05281839519739151
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07368957996368408,
      "epoch": 8.5,
      "learning_rate": 4.459551289459684e-05,
      "loss": 0.0714,
      "step": 8948,
      "task_loss": 0.05031800642609596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06434916704893112,
      "epoch": 8.5,
      "learning_rate": 4.4588894341227426e-05,
      "loss": 0.0588,
      "step": 8949,
      "task_loss": 0.00913977436721325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04223328083753586,
      "epoch": 8.5,
      "learning_rate": 4.4582272229441024e-05,
      "loss": 0.0406,
      "step": 8950,
      "task_loss": 0.02632623352110386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13263371586799622,
      "epoch": 8.5,
      "learning_rate": 4.457564656044056e-05,
      "loss": 0.1389,
      "step": 8951,
      "task_loss": 0.1952933669090271
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019756224006414413,
      "epoch": 8.5,
      "learning_rate": 4.456901733542962e-05,
      "loss": 0.0185,
      "step": 8952,
      "task_loss": 0.007240481674671173
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1520935297012329,
      "epoch": 8.5,
      "learning_rate": 4.4562384555612436e-05,
      "loss": 0.1599,
      "step": 8953,
      "task_loss": 0.2299913763999939
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10879489779472351,
      "epoch": 8.5,
      "learning_rate": 4.455574822219388e-05,
      "loss": 0.1219,
      "step": 8954,
      "task_loss": 0.23940874636173248
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0421454980969429,
      "epoch": 8.5,
      "learning_rate": 4.454910833637949e-05,
      "loss": 0.0453,
      "step": 8955,
      "task_loss": 0.07326261699199677
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1279737502336502,
      "epoch": 8.51,
      "learning_rate": 4.454246489937541e-05,
      "loss": 0.1322,
      "step": 8956,
      "task_loss": 0.1699049174785614
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1273374855518341,
      "epoch": 8.51,
      "learning_rate": 4.4535817912388466e-05,
      "loss": 0.1269,
      "step": 8957,
      "task_loss": 0.12270753085613251
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03381875157356262,
      "epoch": 8.51,
      "learning_rate": 4.4529167376626116e-05,
      "loss": 0.061,
      "step": 8958,
      "task_loss": 0.3059147894382477
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.035635169595479965,
      "epoch": 8.51,
      "learning_rate": 4.4522513293296456e-05,
      "loss": 0.0395,
      "step": 8959,
      "task_loss": 0.07383869588375092
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08485317975282669,
      "epoch": 8.51,
      "learning_rate": 4.451585566360823e-05,
      "loss": 0.0817,
      "step": 8960,
      "task_loss": 0.053764086216688156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.041285235434770584,
      "epoch": 8.51,
      "learning_rate": 4.450919448877084e-05,
      "loss": 0.0389,
      "step": 8961,
      "task_loss": 0.017382560297846794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01698409393429756,
      "epoch": 8.51,
      "learning_rate": 4.4502529769994314e-05,
      "loss": 0.0262,
      "step": 8962,
      "task_loss": 0.10962288081645966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.15635554492473602,
      "epoch": 8.51,
      "learning_rate": 4.449586150848934e-05,
      "loss": 0.157,
      "step": 8963,
      "task_loss": 0.16323071718215942
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016910862177610397,
      "epoch": 8.51,
      "learning_rate": 4.44891897054672e-05,
      "loss": 0.0221,
      "step": 8964,
      "task_loss": 0.06874377280473709
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018827490508556366,
      "epoch": 8.51,
      "learning_rate": 4.4482514362139915e-05,
      "loss": 0.0173,
      "step": 8965,
      "task_loss": 0.003253905102610588
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.044889599084854126,
      "epoch": 8.51,
      "learning_rate": 4.4475835479720065e-05,
      "loss": 0.0419,
      "step": 8966,
      "task_loss": 0.014503007754683495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027362622320652008,
      "epoch": 8.52,
      "learning_rate": 4.4469153059420895e-05,
      "loss": 0.0301,
      "step": 8967,
      "task_loss": 0.054887376725673676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021328095346689224,
      "epoch": 8.52,
      "learning_rate": 4.4462467102456305e-05,
      "loss": 0.0198,
      "step": 8968,
      "task_loss": 0.005754020065069199
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03476056456565857,
      "epoch": 8.52,
      "learning_rate": 4.4455777610040846e-05,
      "loss": 0.0323,
      "step": 8969,
      "task_loss": 0.010109812021255493
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09709491580724716,
      "epoch": 8.52,
      "learning_rate": 4.444908458338968e-05,
      "loss": 0.0945,
      "step": 8970,
      "task_loss": 0.07128534466028214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.037777457386255264,
      "epoch": 8.52,
      "learning_rate": 4.4442388023718624e-05,
      "loss": 0.0562,
      "step": 8971,
      "task_loss": 0.2219763994216919
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05716599524021149,
      "epoch": 8.52,
      "learning_rate": 4.443568793224415e-05,
      "loss": 0.0549,
      "step": 8972,
      "task_loss": 0.03474997729063034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03178703784942627,
      "epoch": 8.52,
      "learning_rate": 4.4428984310183364e-05,
      "loss": 0.0331,
      "step": 8973,
      "task_loss": 0.04533195123076439
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06467156112194061,
      "epoch": 8.52,
      "learning_rate": 4.4422277158754005e-05,
      "loss": 0.0693,
      "step": 8974,
      "task_loss": 0.11103697121143341
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1044667512178421,
      "epoch": 8.52,
      "learning_rate": 4.441556647917446e-05,
      "loss": 0.106,
      "step": 8975,
      "task_loss": 0.11955499649047852
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06567876040935516,
      "epoch": 8.52,
      "learning_rate": 4.440885227266376e-05,
      "loss": 0.0679,
      "step": 8976,
      "task_loss": 0.08836454898118973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09253351390361786,
      "epoch": 8.53,
      "learning_rate": 4.440213454044158e-05,
      "loss": 0.0892,
      "step": 8977,
      "task_loss": 0.05876833572983742
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02796206995844841,
      "epoch": 8.53,
      "learning_rate": 4.43954132837282e-05,
      "loss": 0.0414,
      "step": 8978,
      "task_loss": 0.16239336133003235
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07454772293567657,
      "epoch": 8.53,
      "learning_rate": 4.43886885037446e-05,
      "loss": 0.0779,
      "step": 8979,
      "task_loss": 0.10792988538742065
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08012815564870834,
      "epoch": 8.53,
      "learning_rate": 4.438196020171235e-05,
      "loss": 0.0856,
      "step": 8980,
      "task_loss": 0.13467732071876526
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06739650666713715,
      "epoch": 8.53,
      "learning_rate": 4.437522837885369e-05,
      "loss": 0.0743,
      "step": 8981,
      "task_loss": 0.13656720519065857
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.039198122918605804,
      "epoch": 8.53,
      "learning_rate": 4.436849303639148e-05,
      "loss": 0.0449,
      "step": 8982,
      "task_loss": 0.09616616368293762
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02301427721977234,
      "epoch": 8.53,
      "learning_rate": 4.436175417554923e-05,
      "loss": 0.0213,
      "step": 8983,
      "task_loss": 0.005967108532786369
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025122862309217453,
      "epoch": 8.53,
      "learning_rate": 4.4355011797551086e-05,
      "loss": 0.0323,
      "step": 8984,
      "task_loss": 0.09659717977046967
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06348798424005508,
      "epoch": 8.53,
      "learning_rate": 4.4348265903621844e-05,
      "loss": 0.0593,
      "step": 8985,
      "task_loss": 0.02176908776164055
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024741999804973602,
      "epoch": 8.53,
      "learning_rate": 4.4341516494986904e-05,
      "loss": 0.029,
      "step": 8986,
      "task_loss": 0.06750153005123138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022810565307736397,
      "epoch": 8.53,
      "learning_rate": 4.433476357287235e-05,
      "loss": 0.029,
      "step": 8987,
      "task_loss": 0.08470144867897034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03231749311089516,
      "epoch": 8.54,
      "learning_rate": 4.432800713850488e-05,
      "loss": 0.04,
      "step": 8988,
      "task_loss": 0.10918974876403809
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08315695822238922,
      "epoch": 8.54,
      "learning_rate": 4.432124719311182e-05,
      "loss": 0.0804,
      "step": 8989,
      "task_loss": 0.056007783859968185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07280036807060242,
      "epoch": 8.54,
      "learning_rate": 4.431448373792116e-05,
      "loss": 0.0704,
      "step": 8990,
      "task_loss": 0.048897065222263336
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023964036256074905,
      "epoch": 8.54,
      "learning_rate": 4.430771677416151e-05,
      "loss": 0.0321,
      "step": 8991,
      "task_loss": 0.10551959276199341
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012696181423962116,
      "epoch": 8.54,
      "learning_rate": 4.430094630306212e-05,
      "loss": 0.0121,
      "step": 8992,
      "task_loss": 0.006786322221159935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06884432584047318,
      "epoch": 8.54,
      "learning_rate": 4.429417232585288e-05,
      "loss": 0.0656,
      "step": 8993,
      "task_loss": 0.036455415189266205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03897576034069061,
      "epoch": 8.54,
      "learning_rate": 4.428739484376431e-05,
      "loss": 0.0436,
      "step": 8994,
      "task_loss": 0.08556883037090302
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018597451969981194,
      "epoch": 8.54,
      "learning_rate": 4.4280613858027584e-05,
      "loss": 0.0173,
      "step": 8995,
      "task_loss": 0.005144355818629265
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02616756409406662,
      "epoch": 8.54,
      "learning_rate": 4.427382936987449e-05,
      "loss": 0.0291,
      "step": 8996,
      "task_loss": 0.05589302256703377
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02272038534283638,
      "epoch": 8.54,
      "learning_rate": 4.426704138053747e-05,
      "loss": 0.0209,
      "step": 8997,
      "task_loss": 0.004764288663864136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017667189240455627,
      "epoch": 8.55,
      "learning_rate": 4.426024989124959e-05,
      "loss": 0.0162,
      "step": 8998,
      "task_loss": 0.00283648818731308
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09155499935150146,
      "epoch": 8.55,
      "learning_rate": 4.425345490324456e-05,
      "loss": 0.0864,
      "step": 8999,
      "task_loss": 0.04047724977135658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0898907482624054,
      "epoch": 8.55,
      "learning_rate": 4.424665641775673e-05,
      "loss": 0.0954,
      "step": 9000,
      "task_loss": 0.14472734928131104
    },
    {
      "epoch": 8.55,
      "eval_accuracy": 0.893348623853211,
      "eval_loss": 0.4816896319389343,
      "eval_runtime": 18.2944,
      "eval_samples_per_second": 47.665,
      "eval_steps_per_second": 5.958,
      "step": 9000
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1226925477385521,
      "epoch": 8.55,
      "learning_rate": 4.4239854436021056e-05,
      "loss": 0.1208,
      "step": 9001,
      "task_loss": 0.10388394445180893
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07912658900022507,
      "epoch": 8.55,
      "learning_rate": 4.423304895927317e-05,
      "loss": 0.0727,
      "step": 9002,
      "task_loss": 0.014653431251645088
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07628723978996277,
      "epoch": 8.55,
      "learning_rate": 4.4226239988749305e-05,
      "loss": 0.0717,
      "step": 9003,
      "task_loss": 0.030388563871383667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.17722564935684204,
      "epoch": 8.55,
      "learning_rate": 4.4219427525686366e-05,
      "loss": 0.1661,
      "step": 9004,
      "task_loss": 0.06582056730985641
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01308794878423214,
      "epoch": 8.55,
      "learning_rate": 4.421261157132185e-05,
      "loss": 0.0245,
      "step": 9005,
      "task_loss": 0.1268969178199768
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06603103131055832,
      "epoch": 8.55,
      "learning_rate": 4.4205792126893905e-05,
      "loss": 0.0753,
      "step": 9006,
      "task_loss": 0.15897931158542633
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.033943966031074524,
      "epoch": 8.55,
      "learning_rate": 4.4198969193641324e-05,
      "loss": 0.0317,
      "step": 9007,
      "task_loss": 0.01145954243838787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09488087147474289,
      "epoch": 8.55,
      "learning_rate": 4.4192142772803535e-05,
      "loss": 0.0931,
      "step": 9008,
      "task_loss": 0.07663992047309875
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04525774344801903,
      "epoch": 8.56,
      "learning_rate": 4.4185312865620575e-05,
      "loss": 0.0469,
      "step": 9009,
      "task_loss": 0.06158378720283508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022823676466941833,
      "epoch": 8.56,
      "learning_rate": 4.417847947333314e-05,
      "loss": 0.0212,
      "step": 9010,
      "task_loss": 0.006658636033535004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017958180978894234,
      "epoch": 8.56,
      "learning_rate": 4.417164259718254e-05,
      "loss": 0.0167,
      "step": 9011,
      "task_loss": 0.005114752799272537
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12965308129787445,
      "epoch": 8.56,
      "learning_rate": 4.416480223841073e-05,
      "loss": 0.1233,
      "step": 9012,
      "task_loss": 0.06576710194349289
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08245688676834106,
      "epoch": 8.56,
      "learning_rate": 4.4157958398260294e-05,
      "loss": 0.0786,
      "step": 9013,
      "task_loss": 0.04379882290959358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012385990470647812,
      "epoch": 8.56,
      "learning_rate": 4.415111107797445e-05,
      "loss": 0.0117,
      "step": 9014,
      "task_loss": 0.005246100947260857
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08853226900100708,
      "epoch": 8.56,
      "learning_rate": 4.414426027879705e-05,
      "loss": 0.0933,
      "step": 9015,
      "task_loss": 0.13666236400604248
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06031012162566185,
      "epoch": 8.56,
      "learning_rate": 4.413740600197257e-05,
      "loss": 0.0607,
      "step": 9016,
      "task_loss": 0.06431614607572556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.044511131942272186,
      "epoch": 8.56,
      "learning_rate": 4.413054824874612e-05,
      "loss": 0.062,
      "step": 9017,
      "task_loss": 0.21939462423324585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.039623238146305084,
      "epoch": 8.56,
      "learning_rate": 4.412368702036345e-05,
      "loss": 0.0482,
      "step": 9018,
      "task_loss": 0.12542547285556793
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.2884577512741089,
      "epoch": 8.57,
      "learning_rate": 4.4116822318070925e-05,
      "loss": 0.2897,
      "step": 9019,
      "task_loss": 0.3008236885070801
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.044239118695259094,
      "epoch": 8.57,
      "learning_rate": 4.4109954143115565e-05,
      "loss": 0.0403,
      "step": 9020,
      "task_loss": 0.004376189783215523
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01869271881878376,
      "epoch": 8.57,
      "learning_rate": 4.4103082496745e-05,
      "loss": 0.0232,
      "step": 9021,
      "task_loss": 0.06380611658096313
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0486907958984375,
      "epoch": 8.57,
      "learning_rate": 4.40962073802075e-05,
      "loss": 0.0468,
      "step": 9022,
      "task_loss": 0.03009847365319729
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01841585896909237,
      "epoch": 8.57,
      "learning_rate": 4.4089328794751954e-05,
      "loss": 0.0171,
      "step": 9023,
      "task_loss": 0.005129978060722351
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0398845374584198,
      "epoch": 8.57,
      "learning_rate": 4.4082446741627906e-05,
      "loss": 0.0362,
      "step": 9024,
      "task_loss": 0.003042936325073242
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.21794645488262177,
      "epoch": 8.57,
      "learning_rate": 4.40755612220855e-05,
      "loss": 0.2127,
      "step": 9025,
      "task_loss": 0.165152907371521
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07094389200210571,
      "epoch": 8.57,
      "learning_rate": 4.406867223737553e-05,
      "loss": 0.0752,
      "step": 9026,
      "task_loss": 0.11325624585151672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10760428756475449,
      "epoch": 8.57,
      "learning_rate": 4.406177978874941e-05,
      "loss": 0.1094,
      "step": 9027,
      "task_loss": 0.1256953477859497
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07674893736839294,
      "epoch": 8.57,
      "learning_rate": 4.405488387745919e-05,
      "loss": 0.0779,
      "step": 9028,
      "task_loss": 0.08849011361598969
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05093570053577423,
      "epoch": 8.57,
      "learning_rate": 4.4047984504757544e-05,
      "loss": 0.0768,
      "step": 9029,
      "task_loss": 0.3099355101585388
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02506938949227333,
      "epoch": 8.58,
      "learning_rate": 4.4041081671897775e-05,
      "loss": 0.0271,
      "step": 9030,
      "task_loss": 0.04571164399385452
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018875867128372192,
      "epoch": 8.58,
      "learning_rate": 4.403417538013382e-05,
      "loss": 0.0319,
      "step": 9031,
      "task_loss": 0.14952301979064941
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02852119505405426,
      "epoch": 8.58,
      "learning_rate": 4.402726563072024e-05,
      "loss": 0.0358,
      "step": 9032,
      "task_loss": 0.10161833465099335
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022057583555579185,
      "epoch": 8.58,
      "learning_rate": 4.4020352424912226e-05,
      "loss": 0.0208,
      "step": 9033,
      "task_loss": 0.009751364588737488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009804684668779373,
      "epoch": 8.58,
      "learning_rate": 4.401343576396558e-05,
      "loss": 0.0092,
      "step": 9034,
      "task_loss": 0.0038125887513160706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05322907119989395,
      "epoch": 8.58,
      "learning_rate": 4.400651564913676e-05,
      "loss": 0.053,
      "step": 9035,
      "task_loss": 0.05118084326386452
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01981445588171482,
      "epoch": 8.58,
      "learning_rate": 4.399959208168284e-05,
      "loss": 0.0281,
      "step": 9036,
      "task_loss": 0.10224159061908722
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03673945739865303,
      "epoch": 8.58,
      "learning_rate": 4.3992665062861514e-05,
      "loss": 0.0404,
      "step": 9037,
      "task_loss": 0.07305504381656647
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11148297786712646,
      "epoch": 8.58,
      "learning_rate": 4.398573459393111e-05,
      "loss": 0.1081,
      "step": 9038,
      "task_loss": 0.07762963324785233
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11146679520606995,
      "epoch": 8.58,
      "learning_rate": 4.3978800676150575e-05,
      "loss": 0.108,
      "step": 9039,
      "task_loss": 0.0765286237001419
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03709033504128456,
      "epoch": 8.58,
      "learning_rate": 4.39718633107795e-05,
      "loss": 0.0438,
      "step": 9040,
      "task_loss": 0.10385788232088089
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.038635239005088806,
      "epoch": 8.59,
      "learning_rate": 4.3964922499078084e-05,
      "loss": 0.0372,
      "step": 9041,
      "task_loss": 0.02382836863398552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.051188208162784576,
      "epoch": 8.59,
      "learning_rate": 4.3957978242307166e-05,
      "loss": 0.0502,
      "step": 9042,
      "task_loss": 0.04129528999328613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028153028339147568,
      "epoch": 8.59,
      "learning_rate": 4.395103054172819e-05,
      "loss": 0.0258,
      "step": 9043,
      "task_loss": 0.004962790757417679
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022645195946097374,
      "epoch": 8.59,
      "learning_rate": 4.394407939860325e-05,
      "loss": 0.0212,
      "step": 9044,
      "task_loss": 0.008643986657261848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013338697142899036,
      "epoch": 8.59,
      "learning_rate": 4.3937124814195054e-05,
      "loss": 0.0125,
      "step": 9045,
      "task_loss": 0.004968065768480301
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05772208422422409,
      "epoch": 8.59,
      "learning_rate": 4.393016678976692e-05,
      "loss": 0.0609,
      "step": 9046,
      "task_loss": 0.08959686756134033
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03842834755778313,
      "epoch": 8.59,
      "learning_rate": 4.3923205326582837e-05,
      "loss": 0.0449,
      "step": 9047,
      "task_loss": 0.10268554091453552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03277190029621124,
      "epoch": 8.59,
      "learning_rate": 4.3916240425907364e-05,
      "loss": 0.0405,
      "step": 9048,
      "task_loss": 0.11014024913311005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020059367641806602,
      "epoch": 8.59,
      "learning_rate": 4.3909272089005714e-05,
      "loss": 0.0194,
      "step": 9049,
      "task_loss": 0.01322341151535511
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.2156301587820053,
      "epoch": 8.59,
      "learning_rate": 4.3902300317143726e-05,
      "loss": 0.2129,
      "step": 9050,
      "task_loss": 0.18819431960582733
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02864878438413143,
      "epoch": 8.6,
      "learning_rate": 4.389532511158785e-05,
      "loss": 0.0264,
      "step": 9051,
      "task_loss": 0.006369665265083313
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04067389667034149,
      "epoch": 8.6,
      "learning_rate": 4.388834647360516e-05,
      "loss": 0.0411,
      "step": 9052,
      "task_loss": 0.045049235224723816
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019816577434539795,
      "epoch": 8.6,
      "learning_rate": 4.388136440446337e-05,
      "loss": 0.0244,
      "step": 9053,
      "task_loss": 0.06586334109306335
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018740810453891754,
      "epoch": 8.6,
      "learning_rate": 4.387437890543081e-05,
      "loss": 0.0203,
      "step": 9054,
      "task_loss": 0.0338791161775589
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03478642553091049,
      "epoch": 8.6,
      "learning_rate": 4.3867389977776416e-05,
      "loss": 0.0369,
      "step": 9055,
      "task_loss": 0.056206680834293365
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06599288433790207,
      "epoch": 8.6,
      "learning_rate": 4.3860397622769756e-05,
      "loss": 0.0698,
      "step": 9056,
      "task_loss": 0.10445068776607513
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06809773296117783,
      "epoch": 8.6,
      "learning_rate": 4.3853401841681046e-05,
      "loss": 0.0656,
      "step": 9057,
      "task_loss": 0.04264071583747864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0191783644258976,
      "epoch": 8.6,
      "learning_rate": 4.3846402635781093e-05,
      "loss": 0.0178,
      "step": 9058,
      "task_loss": 0.005517646670341492
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020288830623030663,
      "epoch": 8.6,
      "learning_rate": 4.3839400006341335e-05,
      "loss": 0.0208,
      "step": 9059,
      "task_loss": 0.025135664269328117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02526734210550785,
      "epoch": 8.6,
      "learning_rate": 4.383239395463383e-05,
      "loss": 0.0395,
      "step": 9060,
      "task_loss": 0.16764724254608154
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01698536053299904,
      "epoch": 8.6,
      "learning_rate": 4.382538448193127e-05,
      "loss": 0.0157,
      "step": 9061,
      "task_loss": 0.0042114946991205215
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.17208489775657654,
      "epoch": 8.61,
      "learning_rate": 4.381837158950695e-05,
      "loss": 0.179,
      "step": 9062,
      "task_loss": 0.2413436472415924
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03591246157884598,
      "epoch": 8.61,
      "learning_rate": 4.3811355278634804e-05,
      "loss": 0.0522,
      "step": 9063,
      "task_loss": 0.19905783236026764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03541406989097595,
      "epoch": 8.61,
      "learning_rate": 4.380433555058937e-05,
      "loss": 0.0331,
      "step": 9064,
      "task_loss": 0.011965034529566765
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1491689682006836,
      "epoch": 8.61,
      "learning_rate": 4.379731240664583e-05,
      "loss": 0.163,
      "step": 9065,
      "task_loss": 0.2871720790863037
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015187927521765232,
      "epoch": 8.61,
      "learning_rate": 4.379028584807996e-05,
      "loss": 0.0142,
      "step": 9066,
      "task_loss": 0.005585832521319389
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.031154222786426544,
      "epoch": 8.61,
      "learning_rate": 4.3783255876168165e-05,
      "loss": 0.0329,
      "step": 9067,
      "task_loss": 0.04878431186079979
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12299706786870956,
      "epoch": 8.61,
      "learning_rate": 4.377622249218748e-05,
      "loss": 0.135,
      "step": 9068,
      "task_loss": 0.24260729551315308
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03098423220217228,
      "epoch": 8.61,
      "learning_rate": 4.376918569741556e-05,
      "loss": 0.0388,
      "step": 9069,
      "task_loss": 0.1092497706413269
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02339218370616436,
      "epoch": 8.61,
      "learning_rate": 4.376214549313066e-05,
      "loss": 0.0218,
      "step": 9070,
      "task_loss": 0.007723584771156311
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0801115334033966,
      "epoch": 8.61,
      "learning_rate": 4.375510188061167e-05,
      "loss": 0.0918,
      "step": 9071,
      "task_loss": 0.1967240571975708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0836428701877594,
      "epoch": 8.62,
      "learning_rate": 4.37480548611381e-05,
      "loss": 0.0883,
      "step": 9072,
      "task_loss": 0.1297881305217743
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0943288654088974,
      "epoch": 8.62,
      "learning_rate": 4.374100443599007e-05,
      "loss": 0.0906,
      "step": 9073,
      "task_loss": 0.0567990243434906
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07607637345790863,
      "epoch": 8.62,
      "learning_rate": 4.3733950606448324e-05,
      "loss": 0.0873,
      "step": 9074,
      "task_loss": 0.18788626790046692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11874774098396301,
      "epoch": 8.62,
      "learning_rate": 4.3726893373794234e-05,
      "loss": 0.1183,
      "step": 9075,
      "task_loss": 0.11419327557086945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021572759374976158,
      "epoch": 8.62,
      "learning_rate": 4.3719832739309766e-05,
      "loss": 0.0243,
      "step": 9076,
      "task_loss": 0.04849618673324585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026811379939317703,
      "epoch": 8.62,
      "learning_rate": 4.371276870427753e-05,
      "loss": 0.0278,
      "step": 9077,
      "task_loss": 0.03716592118144035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008828360587358475,
      "epoch": 8.62,
      "learning_rate": 4.3705701269980734e-05,
      "loss": 0.0082,
      "step": 9078,
      "task_loss": 0.0030191540718078613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05028454214334488,
      "epoch": 8.62,
      "learning_rate": 4.369863043770322e-05,
      "loss": 0.0475,
      "step": 9079,
      "task_loss": 0.022545762360095978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.047442540526390076,
      "epoch": 8.62,
      "learning_rate": 4.369155620872943e-05,
      "loss": 0.0435,
      "step": 9080,
      "task_loss": 0.007748594507575035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014092406257987022,
      "epoch": 8.62,
      "learning_rate": 4.3684478584344433e-05,
      "loss": 0.0131,
      "step": 9081,
      "task_loss": 0.004466302692890167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07513783872127533,
      "epoch": 8.62,
      "learning_rate": 4.367739756583392e-05,
      "loss": 0.0833,
      "step": 9082,
      "task_loss": 0.1565927267074585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019305624067783356,
      "epoch": 8.63,
      "learning_rate": 4.367031315448419e-05,
      "loss": 0.018,
      "step": 9083,
      "task_loss": 0.006342671811580658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0106560830026865,
      "epoch": 8.63,
      "learning_rate": 4.366322535158215e-05,
      "loss": 0.014,
      "step": 9084,
      "task_loss": 0.043936390429735184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06862330436706543,
      "epoch": 8.63,
      "learning_rate": 4.3656134158415344e-05,
      "loss": 0.0769,
      "step": 9085,
      "task_loss": 0.15090253949165344
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.032704662531614304,
      "epoch": 8.63,
      "learning_rate": 4.364903957627192e-05,
      "loss": 0.0301,
      "step": 9086,
      "task_loss": 0.006765572354197502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022231170907616615,
      "epoch": 8.63,
      "learning_rate": 4.3641941606440644e-05,
      "loss": 0.0292,
      "step": 9087,
      "task_loss": 0.09189935028553009
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.15758824348449707,
      "epoch": 8.63,
      "learning_rate": 4.36348402502109e-05,
      "loss": 0.1631,
      "step": 9088,
      "task_loss": 0.212388277053833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018683135509490967,
      "epoch": 8.63,
      "learning_rate": 4.3627735508872666e-05,
      "loss": 0.0174,
      "step": 9089,
      "task_loss": 0.0056790560483932495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03444650024175644,
      "epoch": 8.63,
      "learning_rate": 4.362062738371657e-05,
      "loss": 0.0375,
      "step": 9090,
      "task_loss": 0.06486361473798752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.032534707337617874,
      "epoch": 8.63,
      "learning_rate": 4.361351587603384e-05,
      "loss": 0.0591,
      "step": 9091,
      "task_loss": 0.29826819896698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02512388862669468,
      "epoch": 8.63,
      "learning_rate": 4.360640098711629e-05,
      "loss": 0.0234,
      "step": 9092,
      "task_loss": 0.007688479498028755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.053498346358537674,
      "epoch": 8.64,
      "learning_rate": 4.3599282718256406e-05,
      "loss": 0.0579,
      "step": 9093,
      "task_loss": 0.09801323711872101
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018389012664556503,
      "epoch": 8.64,
      "learning_rate": 4.3592161070747233e-05,
      "loss": 0.017,
      "step": 9094,
      "task_loss": 0.004471609368920326
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03994788974523544,
      "epoch": 8.64,
      "learning_rate": 4.358503604588247e-05,
      "loss": 0.054,
      "step": 9095,
      "task_loss": 0.1802259087562561
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0729394480586052,
      "epoch": 8.64,
      "learning_rate": 4.357790764495639e-05,
      "loss": 0.0712,
      "step": 9096,
      "task_loss": 0.055580612272024155
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05639396235346794,
      "epoch": 8.64,
      "learning_rate": 4.357077586926392e-05,
      "loss": 0.065,
      "step": 9097,
      "task_loss": 0.1426202952861786
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06405860185623169,
      "epoch": 8.64,
      "learning_rate": 4.356364072010059e-05,
      "loss": 0.0699,
      "step": 9098,
      "task_loss": 0.1220637708902359
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0317995660007,
      "epoch": 8.64,
      "learning_rate": 4.3556502198762496e-05,
      "loss": 0.0292,
      "step": 9099,
      "task_loss": 0.006273902952671051
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11453507095575333,
      "epoch": 8.64,
      "learning_rate": 4.354936030654642e-05,
      "loss": 0.1091,
      "step": 9100,
      "task_loss": 0.060647960752248764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020458191633224487,
      "epoch": 8.64,
      "learning_rate": 4.3542215044749705e-05,
      "loss": 0.0258,
      "step": 9101,
      "task_loss": 0.07395413517951965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04494299367070198,
      "epoch": 8.64,
      "learning_rate": 4.3535066414670336e-05,
      "loss": 0.0484,
      "step": 9102,
      "task_loss": 0.0797007754445076
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04942406713962555,
      "epoch": 8.64,
      "learning_rate": 4.352791441760687e-05,
      "loss": 0.0658,
      "step": 9103,
      "task_loss": 0.21325786411762238
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07142602652311325,
      "epoch": 8.65,
      "learning_rate": 4.352075905485854e-05,
      "loss": 0.0679,
      "step": 9104,
      "task_loss": 0.03596463054418564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03519432991743088,
      "epoch": 8.65,
      "learning_rate": 4.351360032772512e-05,
      "loss": 0.0375,
      "step": 9105,
      "task_loss": 0.05811901390552521
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09797616302967072,
      "epoch": 8.65,
      "learning_rate": 4.3506438237507033e-05,
      "loss": 0.0951,
      "step": 9106,
      "task_loss": 0.06876572221517563
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0369865708053112,
      "epoch": 8.65,
      "learning_rate": 4.3499272785505316e-05,
      "loss": 0.0636,
      "step": 9107,
      "task_loss": 0.3026273250579834
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05089612305164337,
      "epoch": 8.65,
      "learning_rate": 4.349210397302161e-05,
      "loss": 0.0513,
      "step": 9108,
      "task_loss": 0.05530446022748947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019257325679063797,
      "epoch": 8.65,
      "learning_rate": 4.348493180135815e-05,
      "loss": 0.0181,
      "step": 9109,
      "task_loss": 0.007393643260002136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01597008667886257,
      "epoch": 8.65,
      "learning_rate": 4.347775627181782e-05,
      "loss": 0.0151,
      "step": 9110,
      "task_loss": 0.006781516596674919
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05022025108337402,
      "epoch": 8.65,
      "learning_rate": 4.3470577385704056e-05,
      "loss": 0.0523,
      "step": 9111,
      "task_loss": 0.07132594287395477
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07263205945491791,
      "epoch": 8.65,
      "learning_rate": 4.346339514432096e-05,
      "loss": 0.0749,
      "step": 9112,
      "task_loss": 0.09521390497684479
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03168325126171112,
      "epoch": 8.65,
      "learning_rate": 4.345620954897322e-05,
      "loss": 0.0433,
      "step": 9113,
      "task_loss": 0.1476568728685379
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06712833046913147,
      "epoch": 8.66,
      "learning_rate": 4.344902060096612e-05,
      "loss": 0.0705,
      "step": 9114,
      "task_loss": 0.10096494853496552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04527723044157028,
      "epoch": 8.66,
      "learning_rate": 4.344182830160558e-05,
      "loss": 0.0446,
      "step": 9115,
      "task_loss": 0.03821223974227905
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.035746797919273376,
      "epoch": 8.66,
      "learning_rate": 4.343463265219811e-05,
      "loss": 0.0337,
      "step": 9116,
      "task_loss": 0.014950959011912346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0270676426589489,
      "epoch": 8.66,
      "learning_rate": 4.342743365405084e-05,
      "loss": 0.0262,
      "step": 9117,
      "task_loss": 0.018197346478700638
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05599695444107056,
      "epoch": 8.66,
      "learning_rate": 4.3420231308471496e-05,
      "loss": 0.0586,
      "step": 9118,
      "task_loss": 0.08177006244659424
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02762037329375744,
      "epoch": 8.66,
      "learning_rate": 4.3413025616768424e-05,
      "loss": 0.0306,
      "step": 9119,
      "task_loss": 0.057342953979969025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05162864178419113,
      "epoch": 8.66,
      "learning_rate": 4.340581658025058e-05,
      "loss": 0.0553,
      "step": 9120,
      "task_loss": 0.08785432577133179
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029456917196512222,
      "epoch": 8.66,
      "learning_rate": 4.33986042002275e-05,
      "loss": 0.0385,
      "step": 9121,
      "task_loss": 0.11994407325983047
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0898047387599945,
      "epoch": 8.66,
      "learning_rate": 4.339138847800936e-05,
      "loss": 0.105,
      "step": 9122,
      "task_loss": 0.24178752303123474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09941445291042328,
      "epoch": 8.66,
      "learning_rate": 4.3384169414906925e-05,
      "loss": 0.107,
      "step": 9123,
      "task_loss": 0.1748739629983902
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01261752750724554,
      "epoch": 8.66,
      "learning_rate": 4.3376947012231586e-05,
      "loss": 0.0193,
      "step": 9124,
      "task_loss": 0.07987774163484573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02077942155301571,
      "epoch": 8.67,
      "learning_rate": 4.336972127129532e-05,
      "loss": 0.0293,
      "step": 9125,
      "task_loss": 0.10643217712640762
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.16735947132110596,
      "epoch": 8.67,
      "learning_rate": 4.3362492193410705e-05,
      "loss": 0.1573,
      "step": 9126,
      "task_loss": 0.06678342819213867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030655119568109512,
      "epoch": 8.67,
      "learning_rate": 4.335525977989095e-05,
      "loss": 0.0421,
      "step": 9127,
      "task_loss": 0.14501135051250458
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0633954331278801,
      "epoch": 8.67,
      "learning_rate": 4.334802403204986e-05,
      "loss": 0.0723,
      "step": 9128,
      "task_loss": 0.15235358476638794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021779123693704605,
      "epoch": 8.67,
      "learning_rate": 4.334078495120184e-05,
      "loss": 0.0288,
      "step": 9129,
      "task_loss": 0.09162303805351257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11775536835193634,
      "epoch": 8.67,
      "learning_rate": 4.33335425386619e-05,
      "loss": 0.1184,
      "step": 9130,
      "task_loss": 0.1240459531545639
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1136549636721611,
      "epoch": 8.67,
      "learning_rate": 4.332629679574566e-05,
      "loss": 0.1147,
      "step": 9131,
      "task_loss": 0.12449486553668976
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0809292122721672,
      "epoch": 8.67,
      "learning_rate": 4.331904772376935e-05,
      "loss": 0.0773,
      "step": 9132,
      "task_loss": 0.04456901177763939
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02589108981192112,
      "epoch": 8.67,
      "learning_rate": 4.3311795324049795e-05,
      "loss": 0.024,
      "step": 9133,
      "task_loss": 0.006641261279582977
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027674835175275803,
      "epoch": 8.67,
      "learning_rate": 4.3304539597904435e-05,
      "loss": 0.0262,
      "step": 9134,
      "task_loss": 0.013046126812696457
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024105960503220558,
      "epoch": 8.68,
      "learning_rate": 4.3297280546651295e-05,
      "loss": 0.0225,
      "step": 9135,
      "task_loss": 0.007911447435617447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03654143214225769,
      "epoch": 8.68,
      "learning_rate": 4.329001817160903e-05,
      "loss": 0.0334,
      "step": 9136,
      "task_loss": 0.0051150210201740265
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07622884958982468,
      "epoch": 8.68,
      "learning_rate": 4.3282752474096864e-05,
      "loss": 0.0791,
      "step": 9137,
      "task_loss": 0.10472090542316437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03936569392681122,
      "epoch": 8.68,
      "learning_rate": 4.327548345543467e-05,
      "loss": 0.0384,
      "step": 9138,
      "task_loss": 0.029306685552001
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015473714098334312,
      "epoch": 8.68,
      "learning_rate": 4.326821111694289e-05,
      "loss": 0.0146,
      "step": 9139,
      "task_loss": 0.00628245621919632
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025059562176465988,
      "epoch": 8.68,
      "learning_rate": 4.3260935459942584e-05,
      "loss": 0.0311,
      "step": 9140,
      "task_loss": 0.08575528860092163
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0825490728020668,
      "epoch": 8.68,
      "learning_rate": 4.32536564857554e-05,
      "loss": 0.0862,
      "step": 9141,
      "task_loss": 0.11920452117919922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03180139511823654,
      "epoch": 8.68,
      "learning_rate": 4.3246374195703604e-05,
      "loss": 0.0335,
      "step": 9142,
      "task_loss": 0.04881680756807327
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06249899044632912,
      "epoch": 8.68,
      "learning_rate": 4.3239088591110065e-05,
      "loss": 0.0643,
      "step": 9143,
      "task_loss": 0.08072178065776825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06553536653518677,
      "epoch": 8.68,
      "learning_rate": 4.323179967329824e-05,
      "loss": 0.0638,
      "step": 9144,
      "task_loss": 0.048590727150440216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01741437427699566,
      "epoch": 8.68,
      "learning_rate": 4.3224507443592196e-05,
      "loss": 0.0162,
      "step": 9145,
      "task_loss": 0.005251972004771233
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.15998907387256622,
      "epoch": 8.69,
      "learning_rate": 4.321721190331661e-05,
      "loss": 0.1598,
      "step": 9146,
      "task_loss": 0.15769526362419128
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.041096534579992294,
      "epoch": 8.69,
      "learning_rate": 4.3209913053796746e-05,
      "loss": 0.0477,
      "step": 9147,
      "task_loss": 0.10717198252677917
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06216401606798172,
      "epoch": 8.69,
      "learning_rate": 4.3202610896358474e-05,
      "loss": 0.0581,
      "step": 9148,
      "task_loss": 0.02124890312552452
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02210739627480507,
      "epoch": 8.69,
      "learning_rate": 4.319530543232827e-05,
      "loss": 0.0205,
      "step": 9149,
      "task_loss": 0.006114525720477104
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.049875520169734955,
      "epoch": 8.69,
      "learning_rate": 4.31879966630332e-05,
      "loss": 0.0462,
      "step": 9150,
      "task_loss": 0.012633267790079117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0812768042087555,
      "epoch": 8.69,
      "learning_rate": 4.318068458980095e-05,
      "loss": 0.089,
      "step": 9151,
      "task_loss": 0.15828600525856018
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03615918010473251,
      "epoch": 8.69,
      "learning_rate": 4.317336921395978e-05,
      "loss": 0.0358,
      "step": 9152,
      "task_loss": 0.03290475159883499
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01615232229232788,
      "epoch": 8.69,
      "learning_rate": 4.316605053683856e-05,
      "loss": 0.0159,
      "step": 9153,
      "task_loss": 0.013984838500618935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12520606815814972,
      "epoch": 8.69,
      "learning_rate": 4.3158728559766786e-05,
      "loss": 0.1234,
      "step": 9154,
      "task_loss": 0.10756815969944
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021900830790400505,
      "epoch": 8.69,
      "learning_rate": 4.315140328407451e-05,
      "loss": 0.0242,
      "step": 9155,
      "task_loss": 0.044664304703474045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0630156397819519,
      "epoch": 8.7,
      "learning_rate": 4.314407471109241e-05,
      "loss": 0.0814,
      "step": 9156,
      "task_loss": 0.2466500848531723
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.038144879043102264,
      "epoch": 8.7,
      "learning_rate": 4.313674284215176e-05,
      "loss": 0.0364,
      "step": 9157,
      "task_loss": 0.020709164440631866
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09585539251565933,
      "epoch": 8.7,
      "learning_rate": 4.312940767858441e-05,
      "loss": 0.0947,
      "step": 9158,
      "task_loss": 0.08445055782794952
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021123457700014114,
      "epoch": 8.7,
      "learning_rate": 4.312206922172286e-05,
      "loss": 0.0196,
      "step": 9159,
      "task_loss": 0.005899334326386452
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026478923857212067,
      "epoch": 8.7,
      "learning_rate": 4.311472747290015e-05,
      "loss": 0.0246,
      "step": 9160,
      "task_loss": 0.007890569046139717
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.048453450202941895,
      "epoch": 8.7,
      "learning_rate": 4.310738243344996e-05,
      "loss": 0.0517,
      "step": 9161,
      "task_loss": 0.0805249959230423
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.040751051157712936,
      "epoch": 8.7,
      "learning_rate": 4.310003410470653e-05,
      "loss": 0.0386,
      "step": 9162,
      "task_loss": 0.018803205341100693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04210975766181946,
      "epoch": 8.7,
      "learning_rate": 4.309268248800476e-05,
      "loss": 0.0473,
      "step": 9163,
      "task_loss": 0.09379490464925766
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0314488485455513,
      "epoch": 8.7,
      "learning_rate": 4.3085327584680056e-05,
      "loss": 0.0393,
      "step": 9164,
      "task_loss": 0.10950774699449539
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023545613512396812,
      "epoch": 8.7,
      "learning_rate": 4.3077969396068505e-05,
      "loss": 0.0339,
      "step": 9165,
      "task_loss": 0.12670263648033142
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026402778923511505,
      "epoch": 8.7,
      "learning_rate": 4.307060792350675e-05,
      "loss": 0.0296,
      "step": 9166,
      "task_loss": 0.057920970022678375
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022520121186971664,
      "epoch": 8.71,
      "learning_rate": 4.306324316833203e-05,
      "loss": 0.0211,
      "step": 9167,
      "task_loss": 0.008033214136958122
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02548299916088581,
      "epoch": 8.71,
      "learning_rate": 4.3055875131882204e-05,
      "loss": 0.0301,
      "step": 9168,
      "task_loss": 0.071867436170578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11392084509134293,
      "epoch": 8.71,
      "learning_rate": 4.30485038154957e-05,
      "loss": 0.1113,
      "step": 9169,
      "task_loss": 0.08807753771543503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013116149231791496,
      "epoch": 8.71,
      "learning_rate": 4.304112922051155e-05,
      "loss": 0.0123,
      "step": 9170,
      "task_loss": 0.005404811352491379
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09974595159292221,
      "epoch": 8.71,
      "learning_rate": 4.30337513482694e-05,
      "loss": 0.0927,
      "step": 9171,
      "task_loss": 0.028922712430357933
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017070695757865906,
      "epoch": 8.71,
      "learning_rate": 4.3026370200109463e-05,
      "loss": 0.0241,
      "step": 9172,
      "task_loss": 0.08706965297460556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04366880655288696,
      "epoch": 8.71,
      "learning_rate": 4.301898577737255e-05,
      "loss": 0.0453,
      "step": 9173,
      "task_loss": 0.06001284718513489
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021489912644028664,
      "epoch": 8.71,
      "learning_rate": 4.3011598081400105e-05,
      "loss": 0.0281,
      "step": 9174,
      "task_loss": 0.08786029368638992
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02259124629199505,
      "epoch": 8.71,
      "learning_rate": 4.3004207113534124e-05,
      "loss": 0.0246,
      "step": 9175,
      "task_loss": 0.04274073615670204
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07785982638597488,
      "epoch": 8.71,
      "learning_rate": 4.2996812875117206e-05,
      "loss": 0.0815,
      "step": 9176,
      "task_loss": 0.114007368683815
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02123086154460907,
      "epoch": 8.72,
      "learning_rate": 4.2989415367492556e-05,
      "loss": 0.02,
      "step": 9177,
      "task_loss": 0.008761711418628693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020061947405338287,
      "epoch": 8.72,
      "learning_rate": 4.298201459200397e-05,
      "loss": 0.0188,
      "step": 9178,
      "task_loss": 0.007334306836128235
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04284045845270157,
      "epoch": 8.72,
      "learning_rate": 4.2974610549995834e-05,
      "loss": 0.041,
      "step": 9179,
      "task_loss": 0.024930864572525024
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05956036597490311,
      "epoch": 8.72,
      "learning_rate": 4.296720324281311e-05,
      "loss": 0.0714,
      "step": 9180,
      "task_loss": 0.17790013551712036
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.035681623965501785,
      "epoch": 8.72,
      "learning_rate": 4.29597926718014e-05,
      "loss": 0.0331,
      "step": 9181,
      "task_loss": 0.009932447224855423
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03028404340147972,
      "epoch": 8.72,
      "learning_rate": 4.295237883830685e-05,
      "loss": 0.0336,
      "step": 9182,
      "task_loss": 0.06349918246269226
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.056420549750328064,
      "epoch": 8.72,
      "learning_rate": 4.294496174367623e-05,
      "loss": 0.0584,
      "step": 9183,
      "task_loss": 0.0761711597442627
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08292526006698608,
      "epoch": 8.72,
      "learning_rate": 4.2937541389256877e-05,
      "loss": 0.0884,
      "step": 9184,
      "task_loss": 0.13766731321811676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013823905028402805,
      "epoch": 8.72,
      "learning_rate": 4.293011777639675e-05,
      "loss": 0.0128,
      "step": 9185,
      "task_loss": 0.0035786759108304977
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04730577394366264,
      "epoch": 8.72,
      "learning_rate": 4.2922690906444374e-05,
      "loss": 0.0534,
      "step": 9186,
      "task_loss": 0.1083056777715683
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07880745083093643,
      "epoch": 8.72,
      "learning_rate": 4.291526078074888e-05,
      "loss": 0.0767,
      "step": 9187,
      "task_loss": 0.05808747559785843
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027023762464523315,
      "epoch": 8.73,
      "learning_rate": 4.290782740065997e-05,
      "loss": 0.0305,
      "step": 9188,
      "task_loss": 0.061429526656866074
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06073024123907089,
      "epoch": 8.73,
      "learning_rate": 4.290039076752799e-05,
      "loss": 0.0719,
      "step": 9189,
      "task_loss": 0.172482430934906
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015384482219815254,
      "epoch": 8.73,
      "learning_rate": 4.28929508827038e-05,
      "loss": 0.0144,
      "step": 9190,
      "task_loss": 0.00527518056333065
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01384667120873928,
      "epoch": 8.73,
      "learning_rate": 4.288550774753892e-05,
      "loss": 0.0179,
      "step": 9191,
      "task_loss": 0.054292093962430954
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03736741095781326,
      "epoch": 8.73,
      "learning_rate": 4.2878061363385414e-05,
      "loss": 0.0363,
      "step": 9192,
      "task_loss": 0.02659483253955841
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025986768305301666,
      "epoch": 8.73,
      "learning_rate": 4.287061173159597e-05,
      "loss": 0.0431,
      "step": 9193,
      "task_loss": 0.19738051295280457
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02209838107228279,
      "epoch": 8.73,
      "learning_rate": 4.286315885352382e-05,
      "loss": 0.03,
      "step": 9194,
      "task_loss": 0.10124228149652481
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024954846128821373,
      "epoch": 8.73,
      "learning_rate": 4.285570273052285e-05,
      "loss": 0.0387,
      "step": 9195,
      "task_loss": 0.1625334769487381
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025567319244146347,
      "epoch": 8.73,
      "learning_rate": 4.2848243363947484e-05,
      "loss": 0.0244,
      "step": 9196,
      "task_loss": 0.014310002326965332
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0735805556178093,
      "epoch": 8.73,
      "learning_rate": 4.2840780755152746e-05,
      "loss": 0.0714,
      "step": 9197,
      "task_loss": 0.0521991103887558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0836353451013565,
      "epoch": 8.74,
      "learning_rate": 4.283331490549426e-05,
      "loss": 0.0817,
      "step": 9198,
      "task_loss": 0.06436392664909363
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04380929470062256,
      "epoch": 8.74,
      "learning_rate": 4.282584581632824e-05,
      "loss": 0.0602,
      "step": 9199,
      "task_loss": 0.20809507369995117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12274999171495438,
      "epoch": 8.74,
      "learning_rate": 4.281837348901148e-05,
      "loss": 0.126,
      "step": 9200,
      "task_loss": 0.1554453819990158
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.057108841836452484,
      "epoch": 8.74,
      "learning_rate": 4.281089792490136e-05,
      "loss": 0.0531,
      "step": 9201,
      "task_loss": 0.016610583290457726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03790339082479477,
      "epoch": 8.74,
      "learning_rate": 4.280341912535585e-05,
      "loss": 0.0477,
      "step": 9202,
      "task_loss": 0.1361059844493866
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06707513332366943,
      "epoch": 8.74,
      "learning_rate": 4.2795937091733515e-05,
      "loss": 0.076,
      "step": 9203,
      "task_loss": 0.1563674360513687
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06679324805736542,
      "epoch": 8.74,
      "learning_rate": 4.27884518253935e-05,
      "loss": 0.0657,
      "step": 9204,
      "task_loss": 0.056202180683612823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04162095487117767,
      "epoch": 8.74,
      "learning_rate": 4.278096332769555e-05,
      "loss": 0.0453,
      "step": 9205,
      "task_loss": 0.07872645556926727
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028269974514842033,
      "epoch": 8.74,
      "learning_rate": 4.277347159999997e-05,
      "loss": 0.0262,
      "step": 9206,
      "task_loss": 0.00805065967142582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1113031655550003,
      "epoch": 8.74,
      "learning_rate": 4.276597664366767e-05,
      "loss": 0.1132,
      "step": 9207,
      "task_loss": 0.1301630735397339
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028668878600001335,
      "epoch": 8.74,
      "learning_rate": 4.2758478460060166e-05,
      "loss": 0.0301,
      "step": 9208,
      "task_loss": 0.04257269576191902
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03439326956868172,
      "epoch": 8.75,
      "learning_rate": 4.275097705053951e-05,
      "loss": 0.0324,
      "step": 9209,
      "task_loss": 0.014087924733757973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02624572440981865,
      "epoch": 8.75,
      "learning_rate": 4.2743472416468385e-05,
      "loss": 0.0285,
      "step": 9210,
      "task_loss": 0.048966288566589355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013097495771944523,
      "epoch": 8.75,
      "learning_rate": 4.2735964559210054e-05,
      "loss": 0.0122,
      "step": 9211,
      "task_loss": 0.004471452906727791
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08438113331794739,
      "epoch": 8.75,
      "learning_rate": 4.272845348012833e-05,
      "loss": 0.091,
      "step": 9212,
      "task_loss": 0.1502004712820053
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015331685543060303,
      "epoch": 8.75,
      "learning_rate": 4.272093918058766e-05,
      "loss": 0.0143,
      "step": 9213,
      "task_loss": 0.004537465050816536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019342856481671333,
      "epoch": 8.75,
      "learning_rate": 4.271342166195304e-05,
      "loss": 0.0346,
      "step": 9214,
      "task_loss": 0.17235919833183289
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023674393072724342,
      "epoch": 8.75,
      "learning_rate": 4.2705900925590056e-05,
      "loss": 0.0295,
      "step": 9215,
      "task_loss": 0.08143065869808197
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013667159713804722,
      "epoch": 8.75,
      "learning_rate": 4.269837697286491e-05,
      "loss": 0.0127,
      "step": 9216,
      "task_loss": 0.0043479762971401215
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03081374801695347,
      "epoch": 8.75,
      "learning_rate": 4.269084980514434e-05,
      "loss": 0.0283,
      "step": 9217,
      "task_loss": 0.0054970309138298035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019850486889481544,
      "epoch": 8.75,
      "learning_rate": 4.268331942379571e-05,
      "loss": 0.0197,
      "step": 9218,
      "task_loss": 0.018215559422969818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.053914956748485565,
      "epoch": 8.75,
      "learning_rate": 4.267578583018694e-05,
      "loss": 0.0562,
      "step": 9219,
      "task_loss": 0.07634520530700684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0433831587433815,
      "epoch": 8.76,
      "learning_rate": 4.2668249025686545e-05,
      "loss": 0.0427,
      "step": 9220,
      "task_loss": 0.03700166195631027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09998966753482819,
      "epoch": 8.76,
      "learning_rate": 4.2660709011663624e-05,
      "loss": 0.1053,
      "step": 9221,
      "task_loss": 0.1534278392791748
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.17341908812522888,
      "epoch": 8.76,
      "learning_rate": 4.2653165789487864e-05,
      "loss": 0.175,
      "step": 9222,
      "task_loss": 0.189271941781044
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021686669439077377,
      "epoch": 8.76,
      "learning_rate": 4.2645619360529514e-05,
      "loss": 0.0204,
      "step": 9223,
      "task_loss": 0.009309127926826477
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030826609581708908,
      "epoch": 8.76,
      "learning_rate": 4.2638069726159424e-05,
      "loss": 0.0394,
      "step": 9224,
      "task_loss": 0.11687062680721283
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01709115505218506,
      "epoch": 8.76,
      "learning_rate": 4.263051688774902e-05,
      "loss": 0.0219,
      "step": 9225,
      "task_loss": 0.06518833339214325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03491377830505371,
      "epoch": 8.76,
      "learning_rate": 4.262296084667032e-05,
      "loss": 0.0541,
      "step": 9226,
      "task_loss": 0.22655193507671356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0234515480697155,
      "epoch": 8.76,
      "learning_rate": 4.2615401604295905e-05,
      "loss": 0.0265,
      "step": 9227,
      "task_loss": 0.0536535307765007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09627413749694824,
      "epoch": 8.76,
      "learning_rate": 4.260783916199895e-05,
      "loss": 0.0981,
      "step": 9228,
      "task_loss": 0.11438637971878052
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.065280482172966,
      "epoch": 8.76,
      "learning_rate": 4.260027352115321e-05,
      "loss": 0.0611,
      "step": 9229,
      "task_loss": 0.02382420189678669
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04924977943301201,
      "epoch": 8.77,
      "learning_rate": 4.2592704683133035e-05,
      "loss": 0.0536,
      "step": 9230,
      "task_loss": 0.09264830499887466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02826646901667118,
      "epoch": 8.77,
      "learning_rate": 4.258513264931331e-05,
      "loss": 0.0412,
      "step": 9231,
      "task_loss": 0.1579834669828415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018645115196704865,
      "epoch": 8.77,
      "learning_rate": 4.257755742106956e-05,
      "loss": 0.0172,
      "step": 9232,
      "task_loss": 0.0041794683784246445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012779043056070805,
      "epoch": 8.77,
      "learning_rate": 4.256997899977784e-05,
      "loss": 0.012,
      "step": 9233,
      "task_loss": 0.004724707454442978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03663068264722824,
      "epoch": 8.77,
      "learning_rate": 4.2562397386814823e-05,
      "loss": 0.042,
      "step": 9234,
      "task_loss": 0.09023445844650269
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018597885966300964,
      "epoch": 8.77,
      "learning_rate": 4.255481258355773e-05,
      "loss": 0.0174,
      "step": 9235,
      "task_loss": 0.006651686504483223
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1508217453956604,
      "epoch": 8.77,
      "learning_rate": 4.254722459138441e-05,
      "loss": 0.1488,
      "step": 9236,
      "task_loss": 0.1301749348640442
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04240924119949341,
      "epoch": 8.77,
      "learning_rate": 4.253963341167321e-05,
      "loss": 0.0484,
      "step": 9237,
      "task_loss": 0.10239571332931519
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.034557685256004333,
      "epoch": 8.77,
      "learning_rate": 4.253203904580314e-05,
      "loss": 0.0318,
      "step": 9238,
      "task_loss": 0.007468333467841148
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.007893304340541363,
      "epoch": 8.77,
      "learning_rate": 4.252444149515374e-05,
      "loss": 0.0165,
      "step": 9239,
      "task_loss": 0.09393760561943054
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03578212484717369,
      "epoch": 8.77,
      "learning_rate": 4.251684076110514e-05,
      "loss": 0.0509,
      "step": 9240,
      "task_loss": 0.1870957762002945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024300862103700638,
      "epoch": 8.78,
      "learning_rate": 4.250923684503806e-05,
      "loss": 0.0296,
      "step": 9241,
      "task_loss": 0.07725972682237625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.14117558300495148,
      "epoch": 8.78,
      "learning_rate": 4.2501629748333774e-05,
      "loss": 0.1339,
      "step": 9242,
      "task_loss": 0.06846613436937332
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10145477950572968,
      "epoch": 8.78,
      "learning_rate": 4.249401947237417e-05,
      "loss": 0.0969,
      "step": 9243,
      "task_loss": 0.056025002151727676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.059924256056547165,
      "epoch": 8.78,
      "learning_rate": 4.248640601854166e-05,
      "loss": 0.0585,
      "step": 9244,
      "task_loss": 0.045406997203826904
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08220744132995605,
      "epoch": 8.78,
      "learning_rate": 4.247878938821929e-05,
      "loss": 0.0747,
      "step": 9245,
      "task_loss": 0.007002789527177811
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.14508725702762604,
      "epoch": 8.78,
      "learning_rate": 4.247116958279065e-05,
      "loss": 0.1431,
      "step": 9246,
      "task_loss": 0.1252875030040741
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03380056470632553,
      "epoch": 8.78,
      "learning_rate": 4.246354660363991e-05,
      "loss": 0.0311,
      "step": 9247,
      "task_loss": 0.007123725488781929
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03263848274946213,
      "epoch": 8.78,
      "learning_rate": 4.245592045215182e-05,
      "loss": 0.0461,
      "step": 9248,
      "task_loss": 0.16699370741844177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04700472578406334,
      "epoch": 8.78,
      "learning_rate": 4.244829112971172e-05,
      "loss": 0.0644,
      "step": 9249,
      "task_loss": 0.22047923505306244
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13564848899841309,
      "epoch": 8.78,
      "learning_rate": 4.24406586377055e-05,
      "loss": 0.1365,
      "step": 9250,
      "task_loss": 0.14452001452445984
    },
    {
      "epoch": 8.78,
      "eval_accuracy": 0.8876146788990825,
      "eval_loss": 0.4403258264064789,
      "eval_runtime": 18.1936,
      "eval_samples_per_second": 47.929,
      "eval_steps_per_second": 5.991,
      "step": 9250
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11035147309303284,
      "epoch": 8.79,
      "learning_rate": 4.2433022977519645e-05,
      "loss": 0.1127,
      "step": 9251,
      "task_loss": 0.13335135579109192
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1526888608932495,
      "epoch": 8.79,
      "learning_rate": 4.2425384150541206e-05,
      "loss": 0.148,
      "step": 9252,
      "task_loss": 0.10532040894031525
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07202281802892685,
      "epoch": 8.79,
      "learning_rate": 4.2417742158157816e-05,
      "loss": 0.0773,
      "step": 9253,
      "task_loss": 0.12501731514930725
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0764976292848587,
      "epoch": 8.79,
      "learning_rate": 4.2410097001757676e-05,
      "loss": 0.0794,
      "step": 9254,
      "task_loss": 0.10542768239974976
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01943717896938324,
      "epoch": 8.79,
      "learning_rate": 4.2402448682729566e-05,
      "loss": 0.0181,
      "step": 9255,
      "task_loss": 0.005980361253023148
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06404136121273041,
      "epoch": 8.79,
      "learning_rate": 4.2394797202462844e-05,
      "loss": 0.0716,
      "step": 9256,
      "task_loss": 0.13929347693920135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0188496895134449,
      "epoch": 8.79,
      "learning_rate": 4.238714256234744e-05,
      "loss": 0.018,
      "step": 9257,
      "task_loss": 0.010533113032579422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021757036447525024,
      "epoch": 8.79,
      "learning_rate": 4.237948476377385e-05,
      "loss": 0.0224,
      "step": 9258,
      "task_loss": 0.028252597898244858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.038637999445199966,
      "epoch": 8.79,
      "learning_rate": 4.237182380813315e-05,
      "loss": 0.0356,
      "step": 9259,
      "task_loss": 0.008690120652318
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03595548868179321,
      "epoch": 8.79,
      "learning_rate": 4.236415969681699e-05,
      "loss": 0.0449,
      "step": 9260,
      "task_loss": 0.12517251074314117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014416642487049103,
      "epoch": 8.79,
      "learning_rate": 4.23564924312176e-05,
      "loss": 0.0177,
      "step": 9261,
      "task_loss": 0.047183021903038025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08503177762031555,
      "epoch": 8.8,
      "learning_rate": 4.2348822012727765e-05,
      "loss": 0.0902,
      "step": 9262,
      "task_loss": 0.13704237341880798
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.060867466032505035,
      "epoch": 8.8,
      "learning_rate": 4.234114844274086e-05,
      "loss": 0.0737,
      "step": 9263,
      "task_loss": 0.1892949640750885
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04127098247408867,
      "epoch": 8.8,
      "learning_rate": 4.2333471722650826e-05,
      "loss": 0.0375,
      "step": 9264,
      "task_loss": 0.0030923504382371902
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02909199520945549,
      "epoch": 8.8,
      "learning_rate": 4.232579185385217e-05,
      "loss": 0.0267,
      "step": 9265,
      "task_loss": 0.004700042307376862
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13650727272033691,
      "epoch": 8.8,
      "learning_rate": 4.231810883773999e-05,
      "loss": 0.1356,
      "step": 9266,
      "task_loss": 0.1277703493833542
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07909749448299408,
      "epoch": 8.8,
      "learning_rate": 4.231042267570993e-05,
      "loss": 0.0816,
      "step": 9267,
      "task_loss": 0.10437445342540741
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03750952333211899,
      "epoch": 8.8,
      "learning_rate": 4.230273336915822e-05,
      "loss": 0.0398,
      "step": 9268,
      "task_loss": 0.06028630957007408
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02200540341436863,
      "epoch": 8.8,
      "learning_rate": 4.2295040919481664e-05,
      "loss": 0.0203,
      "step": 9269,
      "task_loss": 0.005379866808652878
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1982060968875885,
      "epoch": 8.8,
      "learning_rate": 4.228734532807763e-05,
      "loss": 0.19,
      "step": 9270,
      "task_loss": 0.11627163738012314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03145551681518555,
      "epoch": 8.8,
      "learning_rate": 4.2279646596344067e-05,
      "loss": 0.0289,
      "step": 9271,
      "task_loss": 0.006189312785863876
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02007415145635605,
      "epoch": 8.81,
      "learning_rate": 4.227194472567948e-05,
      "loss": 0.0271,
      "step": 9272,
      "task_loss": 0.09050026535987854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013911131769418716,
      "epoch": 8.81,
      "learning_rate": 4.2264239717482945e-05,
      "loss": 0.0129,
      "step": 9273,
      "task_loss": 0.004193015396595001
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0349629707634449,
      "epoch": 8.81,
      "learning_rate": 4.225653157315412e-05,
      "loss": 0.0387,
      "step": 9274,
      "task_loss": 0.07233209162950516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11082993447780609,
      "epoch": 8.81,
      "learning_rate": 4.224882029409323e-05,
      "loss": 0.1144,
      "step": 9275,
      "task_loss": 0.14643532037734985
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014289605431258678,
      "epoch": 8.81,
      "learning_rate": 4.224110588170106e-05,
      "loss": 0.0134,
      "step": 9276,
      "task_loss": 0.005692243576049805
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0677342414855957,
      "epoch": 8.81,
      "learning_rate": 4.223338833737898e-05,
      "loss": 0.0682,
      "step": 9277,
      "task_loss": 0.07213495671749115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0438992902636528,
      "epoch": 8.81,
      "learning_rate": 4.22256676625289e-05,
      "loss": 0.0473,
      "step": 9278,
      "task_loss": 0.07815498113632202
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13519689440727234,
      "epoch": 8.81,
      "learning_rate": 4.221794385855334e-05,
      "loss": 0.1365,
      "step": 9279,
      "task_loss": 0.14803387224674225
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017305288463830948,
      "epoch": 8.81,
      "learning_rate": 4.221021692685534e-05,
      "loss": 0.0161,
      "step": 9280,
      "task_loss": 0.005480002611875534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.036162495613098145,
      "epoch": 8.81,
      "learning_rate": 4.220248686883857e-05,
      "loss": 0.0348,
      "step": 9281,
      "task_loss": 0.022239340469241142
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02372421696782112,
      "epoch": 8.81,
      "learning_rate": 4.21947536859072e-05,
      "loss": 0.0221,
      "step": 9282,
      "task_loss": 0.007254859432578087
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.040189243853092194,
      "epoch": 8.82,
      "learning_rate": 4.218701737946601e-05,
      "loss": 0.0424,
      "step": 9283,
      "task_loss": 0.062415711581707
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03214951977133751,
      "epoch": 8.82,
      "learning_rate": 4.217927795092034e-05,
      "loss": 0.0379,
      "step": 9284,
      "task_loss": 0.08962158858776093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.041563909500837326,
      "epoch": 8.82,
      "learning_rate": 4.21715354016761e-05,
      "loss": 0.0388,
      "step": 9285,
      "task_loss": 0.01408584788441658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05513744056224823,
      "epoch": 8.82,
      "learning_rate": 4.216378973313976e-05,
      "loss": 0.0519,
      "step": 9286,
      "task_loss": 0.02227054536342621
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03985002636909485,
      "epoch": 8.82,
      "learning_rate": 4.215604094671835e-05,
      "loss": 0.0513,
      "step": 9287,
      "task_loss": 0.15437698364257812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.055905863642692566,
      "epoch": 8.82,
      "learning_rate": 4.214828904381947e-05,
      "loss": 0.0522,
      "step": 9288,
      "task_loss": 0.01845194399356842
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.049979522824287415,
      "epoch": 8.82,
      "learning_rate": 4.21405340258513e-05,
      "loss": 0.054,
      "step": 9289,
      "task_loss": 0.09055973589420319
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04201385751366615,
      "epoch": 8.82,
      "learning_rate": 4.213277589422258e-05,
      "loss": 0.0532,
      "step": 9290,
      "task_loss": 0.1543242484331131
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10822063684463501,
      "epoch": 8.82,
      "learning_rate": 4.21250146503426e-05,
      "loss": 0.106,
      "step": 9291,
      "task_loss": 0.08604384958744049
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02375629171729088,
      "epoch": 8.82,
      "learning_rate": 4.2117250295621235e-05,
      "loss": 0.0341,
      "step": 9292,
      "task_loss": 0.1269487887620926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028724079951643944,
      "epoch": 8.83,
      "learning_rate": 4.210948283146892e-05,
      "loss": 0.027,
      "step": 9293,
      "task_loss": 0.011627469211816788
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023904111236333847,
      "epoch": 8.83,
      "learning_rate": 4.210171225929664e-05,
      "loss": 0.022,
      "step": 9294,
      "task_loss": 0.004593405872583389
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030907439067959785,
      "epoch": 8.83,
      "learning_rate": 4.209393858051598e-05,
      "loss": 0.0352,
      "step": 9295,
      "task_loss": 0.0739751011133194
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04190927371382713,
      "epoch": 8.83,
      "learning_rate": 4.208616179653903e-05,
      "loss": 0.0506,
      "step": 9296,
      "task_loss": 0.12926626205444336
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011166459880769253,
      "epoch": 8.83,
      "learning_rate": 4.207838190877852e-05,
      "loss": 0.0194,
      "step": 9297,
      "task_loss": 0.09343065321445465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07796520739793777,
      "epoch": 8.83,
      "learning_rate": 4.2070598918647683e-05,
      "loss": 0.0892,
      "step": 9298,
      "task_loss": 0.19046609103679657
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13977394998073578,
      "epoch": 8.83,
      "learning_rate": 4.206281282756034e-05,
      "loss": 0.137,
      "step": 9299,
      "task_loss": 0.1115388497710228
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01926058530807495,
      "epoch": 8.83,
      "learning_rate": 4.205502363693087e-05,
      "loss": 0.0178,
      "step": 9300,
      "task_loss": 0.004766935482621193
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014616726897656918,
      "epoch": 8.83,
      "learning_rate": 4.204723134817422e-05,
      "loss": 0.0137,
      "step": 9301,
      "task_loss": 0.005341559648513794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.031305767595767975,
      "epoch": 8.83,
      "learning_rate": 4.2039435962705886e-05,
      "loss": 0.0336,
      "step": 9302,
      "task_loss": 0.054229650646448135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06339511275291443,
      "epoch": 8.83,
      "learning_rate": 4.2031637481941954e-05,
      "loss": 0.0737,
      "step": 9303,
      "task_loss": 0.1668478399515152
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03651703521609306,
      "epoch": 8.84,
      "learning_rate": 4.202383590729905e-05,
      "loss": 0.0422,
      "step": 9304,
      "task_loss": 0.09357205033302307
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03856692090630531,
      "epoch": 8.84,
      "learning_rate": 4.201603124019436e-05,
      "loss": 0.0393,
      "step": 9305,
      "task_loss": 0.0458366759121418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022785726934671402,
      "epoch": 8.84,
      "learning_rate": 4.200822348204565e-05,
      "loss": 0.0277,
      "step": 9306,
      "task_loss": 0.07169666886329651
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.054608479142189026,
      "epoch": 8.84,
      "learning_rate": 4.200041263427123e-05,
      "loss": 0.0635,
      "step": 9307,
      "task_loss": 0.14374345541000366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0937652736902237,
      "epoch": 8.84,
      "learning_rate": 4.199259869828998e-05,
      "loss": 0.0977,
      "step": 9308,
      "task_loss": 0.13320079445838928
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0214654803276062,
      "epoch": 8.84,
      "learning_rate": 4.1984781675521345e-05,
      "loss": 0.0285,
      "step": 9309,
      "task_loss": 0.09145978093147278
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02630656771361828,
      "epoch": 8.84,
      "learning_rate": 4.1976961567385306e-05,
      "loss": 0.0298,
      "step": 9310,
      "task_loss": 0.06128865107893944
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029763180762529373,
      "epoch": 8.84,
      "learning_rate": 4.1969138375302445e-05,
      "loss": 0.0275,
      "step": 9311,
      "task_loss": 0.007424212992191315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013515127822756767,
      "epoch": 8.84,
      "learning_rate": 4.1961312100693874e-05,
      "loss": 0.0312,
      "step": 9312,
      "task_loss": 0.19070707261562347
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009992139413952827,
      "epoch": 8.84,
      "learning_rate": 4.1953482744981274e-05,
      "loss": 0.0093,
      "step": 9313,
      "task_loss": 0.0031170137226581573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08150660246610641,
      "epoch": 8.85,
      "learning_rate": 4.194565030958688e-05,
      "loss": 0.0909,
      "step": 9314,
      "task_loss": 0.17554497718811035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01873527467250824,
      "epoch": 8.85,
      "learning_rate": 4.19378147959335e-05,
      "loss": 0.0211,
      "step": 9315,
      "task_loss": 0.04287352040410042
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.16984689235687256,
      "epoch": 8.85,
      "learning_rate": 4.192997620544449e-05,
      "loss": 0.1605,
      "step": 9316,
      "task_loss": 0.0767621323466301
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06193321570754051,
      "epoch": 8.85,
      "learning_rate": 4.192213453954377e-05,
      "loss": 0.0631,
      "step": 9317,
      "task_loss": 0.07395564019680023
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01895410194993019,
      "epoch": 8.85,
      "learning_rate": 4.19142897996558e-05,
      "loss": 0.0228,
      "step": 9318,
      "task_loss": 0.05743853747844696
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09764152765274048,
      "epoch": 8.85,
      "learning_rate": 4.190644198720563e-05,
      "loss": 0.0953,
      "step": 9319,
      "task_loss": 0.07411730289459229
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02355141192674637,
      "epoch": 8.85,
      "learning_rate": 4.189859110361886e-05,
      "loss": 0.0269,
      "step": 9320,
      "task_loss": 0.05675677955150604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.23243659734725952,
      "epoch": 8.85,
      "learning_rate": 4.189073715032163e-05,
      "loss": 0.2339,
      "step": 9321,
      "task_loss": 0.24723902344703674
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.057536251842975616,
      "epoch": 8.85,
      "learning_rate": 4.188288012874065e-05,
      "loss": 0.0635,
      "step": 9322,
      "task_loss": 0.11727714538574219
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07297425717115402,
      "epoch": 8.85,
      "learning_rate": 4.187502004030318e-05,
      "loss": 0.0841,
      "step": 9323,
      "task_loss": 0.1841675043106079
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.19478213787078857,
      "epoch": 8.85,
      "learning_rate": 4.186715688643705e-05,
      "loss": 0.1956,
      "step": 9324,
      "task_loss": 0.202731192111969
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.16516168415546417,
      "epoch": 8.86,
      "learning_rate": 4.185929066857064e-05,
      "loss": 0.1631,
      "step": 9325,
      "task_loss": 0.14431238174438477
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.14519473910331726,
      "epoch": 8.86,
      "learning_rate": 4.1851421388132886e-05,
      "loss": 0.1572,
      "step": 9326,
      "task_loss": 0.2655932903289795
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04612842947244644,
      "epoch": 8.86,
      "learning_rate": 4.1843549046553284e-05,
      "loss": 0.0491,
      "step": 9327,
      "task_loss": 0.07564530521631241
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.16417290270328522,
      "epoch": 8.86,
      "learning_rate": 4.183567364526186e-05,
      "loss": 0.1742,
      "step": 9328,
      "task_loss": 0.26465079188346863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02931727096438408,
      "epoch": 8.86,
      "learning_rate": 4.182779518568926e-05,
      "loss": 0.0414,
      "step": 9329,
      "task_loss": 0.1504097282886505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05426057428121567,
      "epoch": 8.86,
      "learning_rate": 4.181991366926661e-05,
      "loss": 0.0512,
      "step": 9330,
      "task_loss": 0.02333644963800907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.2061796486377716,
      "epoch": 8.86,
      "learning_rate": 4.181202909742564e-05,
      "loss": 0.2011,
      "step": 9331,
      "task_loss": 0.15504643321037292
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01282942108809948,
      "epoch": 8.86,
      "learning_rate": 4.1804141471598604e-05,
      "loss": 0.026,
      "step": 9332,
      "task_loss": 0.14491677284240723
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.057288218289613724,
      "epoch": 8.86,
      "learning_rate": 4.179625079321836e-05,
      "loss": 0.0536,
      "step": 9333,
      "task_loss": 0.020448647439479828
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06341984868049622,
      "epoch": 8.86,
      "learning_rate": 4.1788357063718254e-05,
      "loss": 0.0675,
      "step": 9334,
      "task_loss": 0.10468359291553497
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05787978321313858,
      "epoch": 8.87,
      "learning_rate": 4.178046028453224e-05,
      "loss": 0.0547,
      "step": 9335,
      "task_loss": 0.026474833488464355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05362321063876152,
      "epoch": 8.87,
      "learning_rate": 4.1772560457094795e-05,
      "loss": 0.0721,
      "step": 9336,
      "task_loss": 0.23888705670833588
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06569310277700424,
      "epoch": 8.87,
      "learning_rate": 4.1764657582840965e-05,
      "loss": 0.07,
      "step": 9337,
      "task_loss": 0.10902517288923264
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1104736477136612,
      "epoch": 8.87,
      "learning_rate": 4.175675166320635e-05,
      "loss": 0.1168,
      "step": 9338,
      "task_loss": 0.17340587079524994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014345163479447365,
      "epoch": 8.87,
      "learning_rate": 4.1748842699627094e-05,
      "loss": 0.0135,
      "step": 9339,
      "task_loss": 0.00562736950814724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03695271164178848,
      "epoch": 8.87,
      "learning_rate": 4.17409306935399e-05,
      "loss": 0.0341,
      "step": 9340,
      "task_loss": 0.008127160370349884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01702256128191948,
      "epoch": 8.87,
      "learning_rate": 4.173301564638201e-05,
      "loss": 0.0247,
      "step": 9341,
      "task_loss": 0.09353862702846527
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01691245101392269,
      "epoch": 8.87,
      "learning_rate": 4.1725097559591256e-05,
      "loss": 0.0158,
      "step": 9342,
      "task_loss": 0.005920737981796265
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03731689229607582,
      "epoch": 8.87,
      "learning_rate": 4.1717176434605967e-05,
      "loss": 0.0357,
      "step": 9343,
      "task_loss": 0.021031979471445084
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0925697386264801,
      "epoch": 8.87,
      "learning_rate": 4.170925227286508e-05,
      "loss": 0.1213,
      "step": 9344,
      "task_loss": 0.38006648421287537
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.00923939235508442,
      "epoch": 8.87,
      "learning_rate": 4.170132507580803e-05,
      "loss": 0.0155,
      "step": 9345,
      "task_loss": 0.07144822180271149
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009474704042077065,
      "epoch": 8.88,
      "learning_rate": 4.1693394844874856e-05,
      "loss": 0.0154,
      "step": 9346,
      "task_loss": 0.06834915280342102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0885670930147171,
      "epoch": 8.88,
      "learning_rate": 4.1685461581506115e-05,
      "loss": 0.1049,
      "step": 9347,
      "task_loss": 0.2523471713066101
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1884203851222992,
      "epoch": 8.88,
      "learning_rate": 4.167752528714291e-05,
      "loss": 0.1863,
      "step": 9348,
      "task_loss": 0.16740942001342773
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013086885213851929,
      "epoch": 8.88,
      "learning_rate": 4.166958596322692e-05,
      "loss": 0.0186,
      "step": 9349,
      "task_loss": 0.06813879311084747
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1107698380947113,
      "epoch": 8.88,
      "learning_rate": 4.1661643611200366e-05,
      "loss": 0.1103,
      "step": 9350,
      "task_loss": 0.10653585195541382
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03403695300221443,
      "epoch": 8.88,
      "learning_rate": 4.1653698232506e-05,
      "loss": 0.0334,
      "step": 9351,
      "task_loss": 0.02738173119723797
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016268638893961906,
      "epoch": 8.88,
      "learning_rate": 4.1645749828587145e-05,
      "loss": 0.0161,
      "step": 9352,
      "task_loss": 0.014955738559365273
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.039960239082574844,
      "epoch": 8.88,
      "learning_rate": 4.1637798400887674e-05,
      "loss": 0.0423,
      "step": 9353,
      "task_loss": 0.06289532035589218
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021891163662075996,
      "epoch": 8.88,
      "learning_rate": 4.162984395085198e-05,
      "loss": 0.0206,
      "step": 9354,
      "task_loss": 0.008558722212910652
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05967186763882637,
      "epoch": 8.88,
      "learning_rate": 4.162188647992506e-05,
      "loss": 0.0628,
      "step": 9355,
      "task_loss": 0.09045903384685516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11606243252754211,
      "epoch": 8.89,
      "learning_rate": 4.161392598955239e-05,
      "loss": 0.1182,
      "step": 9356,
      "task_loss": 0.13704703748226166
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05215151980519295,
      "epoch": 8.89,
      "learning_rate": 4.160596248118007e-05,
      "loss": 0.066,
      "step": 9357,
      "task_loss": 0.19018231332302094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016069339588284492,
      "epoch": 8.89,
      "learning_rate": 4.159799595625468e-05,
      "loss": 0.015,
      "step": 9358,
      "task_loss": 0.005058174952864647
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016761645674705505,
      "epoch": 8.89,
      "learning_rate": 4.159002641622338e-05,
      "loss": 0.0165,
      "step": 9359,
      "task_loss": 0.013712173327803612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03098924085497856,
      "epoch": 8.89,
      "learning_rate": 4.1582053862533895e-05,
      "loss": 0.0378,
      "step": 9360,
      "task_loss": 0.0994289442896843
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1274225413799286,
      "epoch": 8.89,
      "learning_rate": 4.157407829663446e-05,
      "loss": 0.1212,
      "step": 9361,
      "task_loss": 0.06485594809055328
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016975758597254753,
      "epoch": 8.89,
      "learning_rate": 4.1566099719973884e-05,
      "loss": 0.0294,
      "step": 9362,
      "task_loss": 0.1415422558784485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13337287306785583,
      "epoch": 8.89,
      "learning_rate": 4.1558118134001514e-05,
      "loss": 0.1266,
      "step": 9363,
      "task_loss": 0.06562095880508423
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09345261007547379,
      "epoch": 8.89,
      "learning_rate": 4.155013354016723e-05,
      "loss": 0.0891,
      "step": 9364,
      "task_loss": 0.04989039897918701
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06697876751422882,
      "epoch": 8.89,
      "learning_rate": 4.154214593992149e-05,
      "loss": 0.0627,
      "step": 9365,
      "task_loss": 0.02389654330909252
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.15374194085597992,
      "epoch": 8.89,
      "learning_rate": 4.1534155334715264e-05,
      "loss": 0.1484,
      "step": 9366,
      "task_loss": 0.1006234660744667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009657006710767746,
      "epoch": 8.9,
      "learning_rate": 4.15261617260001e-05,
      "loss": 0.0192,
      "step": 9367,
      "task_loss": 0.10532604157924652
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0194817166775465,
      "epoch": 8.9,
      "learning_rate": 4.151816511522807e-05,
      "loss": 0.0183,
      "step": 9368,
      "task_loss": 0.0075190383940935135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018920117989182472,
      "epoch": 8.9,
      "learning_rate": 4.151016550385179e-05,
      "loss": 0.0241,
      "step": 9369,
      "task_loss": 0.07103104889392853
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015537586063146591,
      "epoch": 8.9,
      "learning_rate": 4.150216289332443e-05,
      "loss": 0.0144,
      "step": 9370,
      "task_loss": 0.004283284768462181
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07015059888362885,
      "epoch": 8.9,
      "learning_rate": 4.149415728509971e-05,
      "loss": 0.0735,
      "step": 9371,
      "task_loss": 0.10341081768274307
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.17750681936740875,
      "epoch": 8.9,
      "learning_rate": 4.1486148680631875e-05,
      "loss": 0.189,
      "step": 9372,
      "task_loss": 0.29196876287460327
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03413275629281998,
      "epoch": 8.9,
      "learning_rate": 4.147813708137574e-05,
      "loss": 0.0359,
      "step": 9373,
      "task_loss": 0.051412858068943024
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.044314898550510406,
      "epoch": 8.9,
      "learning_rate": 4.1470122488786645e-05,
      "loss": 0.0408,
      "step": 9374,
      "task_loss": 0.00903802365064621
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03658336400985718,
      "epoch": 8.9,
      "learning_rate": 4.146210490432048e-05,
      "loss": 0.0348,
      "step": 9375,
      "task_loss": 0.018489893525838852
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01839321479201317,
      "epoch": 8.9,
      "learning_rate": 4.1454084329433674e-05,
      "loss": 0.0171,
      "step": 9376,
      "task_loss": 0.005394909530878067
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04376121610403061,
      "epoch": 8.91,
      "learning_rate": 4.144606076558321e-05,
      "loss": 0.0486,
      "step": 9377,
      "task_loss": 0.09205228835344315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02223261632025242,
      "epoch": 8.91,
      "learning_rate": 4.14380342142266e-05,
      "loss": 0.029,
      "step": 9378,
      "task_loss": 0.08994803577661514
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029528465121984482,
      "epoch": 8.91,
      "learning_rate": 4.14300046768219e-05,
      "loss": 0.0361,
      "step": 9379,
      "task_loss": 0.09534671902656555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.038068436086177826,
      "epoch": 8.91,
      "learning_rate": 4.1421972154827724e-05,
      "loss": 0.0398,
      "step": 9380,
      "task_loss": 0.055462270975112915
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02015954628586769,
      "epoch": 8.91,
      "learning_rate": 4.141393664970323e-05,
      "loss": 0.0265,
      "step": 9381,
      "task_loss": 0.0838894322514534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02617390640079975,
      "epoch": 8.91,
      "learning_rate": 4.140589816290808e-05,
      "loss": 0.0296,
      "step": 9382,
      "task_loss": 0.060920245945453644
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010754341259598732,
      "epoch": 8.91,
      "learning_rate": 4.1397856695902535e-05,
      "loss": 0.0171,
      "step": 9383,
      "task_loss": 0.07444259524345398
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.052527911961078644,
      "epoch": 8.91,
      "learning_rate": 4.138981225014733e-05,
      "loss": 0.0651,
      "step": 9384,
      "task_loss": 0.17810016870498657
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.035542987287044525,
      "epoch": 8.91,
      "learning_rate": 4.1381764827103806e-05,
      "loss": 0.033,
      "step": 9385,
      "task_loss": 0.010187897831201553
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009268272668123245,
      "epoch": 8.91,
      "learning_rate": 4.13737144282338e-05,
      "loss": 0.0144,
      "step": 9386,
      "task_loss": 0.06069015711545944
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01877554878592491,
      "epoch": 8.91,
      "learning_rate": 4.1365661054999715e-05,
      "loss": 0.0176,
      "step": 9387,
      "task_loss": 0.007025185972452164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.056551381945610046,
      "epoch": 8.92,
      "learning_rate": 4.1357604708864475e-05,
      "loss": 0.067,
      "step": 9388,
      "task_loss": 0.16118671000003815
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12249293923377991,
      "epoch": 8.92,
      "learning_rate": 4.1349545391291563e-05,
      "loss": 0.1238,
      "step": 9389,
      "task_loss": 0.13515231013298035
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04289904981851578,
      "epoch": 8.92,
      "learning_rate": 4.1341483103745006e-05,
      "loss": 0.0393,
      "step": 9390,
      "task_loss": 0.006975535303354263
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.047507885843515396,
      "epoch": 8.92,
      "learning_rate": 4.133341784768933e-05,
      "loss": 0.0585,
      "step": 9391,
      "task_loss": 0.15751934051513672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03881534934043884,
      "epoch": 8.92,
      "learning_rate": 4.132534962458962e-05,
      "loss": 0.0365,
      "step": 9392,
      "task_loss": 0.01544598676264286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027173154056072235,
      "epoch": 8.92,
      "learning_rate": 4.131727843591155e-05,
      "loss": 0.025,
      "step": 9393,
      "task_loss": 0.00519617460668087
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0583278089761734,
      "epoch": 8.92,
      "learning_rate": 4.130920428312127e-05,
      "loss": 0.0568,
      "step": 9394,
      "task_loss": 0.042698342353105545
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1009674221277237,
      "epoch": 8.92,
      "learning_rate": 4.130112716768548e-05,
      "loss": 0.102,
      "step": 9395,
      "task_loss": 0.11100717633962631
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07862529158592224,
      "epoch": 8.92,
      "learning_rate": 4.129304709107143e-05,
      "loss": 0.0764,
      "step": 9396,
      "task_loss": 0.0560125857591629
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01500142365694046,
      "epoch": 8.92,
      "learning_rate": 4.128496405474691e-05,
      "loss": 0.014,
      "step": 9397,
      "task_loss": 0.005422631278634071
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025359123945236206,
      "epoch": 8.92,
      "learning_rate": 4.127687806018024e-05,
      "loss": 0.0406,
      "step": 9398,
      "task_loss": 0.1776316612958908
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.061490267515182495,
      "epoch": 8.93,
      "learning_rate": 4.1268789108840275e-05,
      "loss": 0.0665,
      "step": 9399,
      "task_loss": 0.11160209774971008
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05927234888076782,
      "epoch": 8.93,
      "learning_rate": 4.126069720219642e-05,
      "loss": 0.0682,
      "step": 9400,
      "task_loss": 0.14884845912456512
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12741845846176147,
      "epoch": 8.93,
      "learning_rate": 4.125260234171861e-05,
      "loss": 0.1217,
      "step": 9401,
      "task_loss": 0.07042922079563141
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020310278981924057,
      "epoch": 8.93,
      "learning_rate": 4.12445045288773e-05,
      "loss": 0.0244,
      "step": 9402,
      "task_loss": 0.06154558062553406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.049872465431690216,
      "epoch": 8.93,
      "learning_rate": 4.123640376514353e-05,
      "loss": 0.0495,
      "step": 9403,
      "task_loss": 0.04623175412416458
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018078003078699112,
      "epoch": 8.93,
      "learning_rate": 4.12283000519888e-05,
      "loss": 0.0304,
      "step": 9404,
      "task_loss": 0.14123806357383728
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013958177529275417,
      "epoch": 8.93,
      "learning_rate": 4.122019339088522e-05,
      "loss": 0.0213,
      "step": 9405,
      "task_loss": 0.08724473416805267
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020791584625840187,
      "epoch": 8.93,
      "learning_rate": 4.121208378330539e-05,
      "loss": 0.0194,
      "step": 9406,
      "task_loss": 0.006587089970707893
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03328198194503784,
      "epoch": 8.93,
      "learning_rate": 4.120397123072246e-05,
      "loss": 0.0314,
      "step": 9407,
      "task_loss": 0.014744751155376434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017374586313962936,
      "epoch": 8.93,
      "learning_rate": 4.119585573461012e-05,
      "loss": 0.016,
      "step": 9408,
      "task_loss": 0.0033780932426452637
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03562571480870247,
      "epoch": 8.94,
      "learning_rate": 4.118773729644258e-05,
      "loss": 0.0412,
      "step": 9409,
      "task_loss": 0.09113264083862305
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.043064821511507034,
      "epoch": 8.94,
      "learning_rate": 4.11796159176946e-05,
      "loss": 0.0501,
      "step": 9410,
      "task_loss": 0.11316752433776855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08931790292263031,
      "epoch": 8.94,
      "learning_rate": 4.117149159984147e-05,
      "loss": 0.0858,
      "step": 9411,
      "task_loss": 0.0536871999502182
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0797976404428482,
      "epoch": 8.94,
      "learning_rate": 4.116336434435901e-05,
      "loss": 0.0779,
      "step": 9412,
      "task_loss": 0.060566820204257965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.057995155453681946,
      "epoch": 8.94,
      "learning_rate": 4.115523415272358e-05,
      "loss": 0.0555,
      "step": 9413,
      "task_loss": 0.033352144062519073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03625299036502838,
      "epoch": 8.94,
      "learning_rate": 4.1147101026412046e-05,
      "loss": 0.0428,
      "step": 9414,
      "task_loss": 0.10129890590906143
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04442795366048813,
      "epoch": 8.94,
      "learning_rate": 4.1138964966901853e-05,
      "loss": 0.0427,
      "step": 9415,
      "task_loss": 0.027060629799962044
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01638917066156864,
      "epoch": 8.94,
      "learning_rate": 4.113082597567095e-05,
      "loss": 0.0203,
      "step": 9416,
      "task_loss": 0.055116184055805206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.041155144572257996,
      "epoch": 8.94,
      "learning_rate": 4.112268405419782e-05,
      "loss": 0.0521,
      "step": 9417,
      "task_loss": 0.1504831463098526
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028086768463253975,
      "epoch": 8.94,
      "learning_rate": 4.1114539203961476e-05,
      "loss": 0.0373,
      "step": 9418,
      "task_loss": 0.12012705206871033
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026071809232234955,
      "epoch": 8.94,
      "learning_rate": 4.110639142644149e-05,
      "loss": 0.0252,
      "step": 9419,
      "task_loss": 0.016939345747232437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024052530527114868,
      "epoch": 8.95,
      "learning_rate": 4.109824072311792e-05,
      "loss": 0.0352,
      "step": 9420,
      "task_loss": 0.13542687892913818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.048306502401828766,
      "epoch": 8.95,
      "learning_rate": 4.10900870954714e-05,
      "loss": 0.047,
      "step": 9421,
      "task_loss": 0.03500698506832123
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0342412106692791,
      "epoch": 8.95,
      "learning_rate": 4.108193054498307e-05,
      "loss": 0.0553,
      "step": 9422,
      "task_loss": 0.24514110386371613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013718321919441223,
      "epoch": 8.95,
      "learning_rate": 4.10737710731346e-05,
      "loss": 0.0214,
      "step": 9423,
      "task_loss": 0.09085668623447418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013474099338054657,
      "epoch": 8.95,
      "learning_rate": 4.106560868140821e-05,
      "loss": 0.0191,
      "step": 9424,
      "task_loss": 0.07011019438505173
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0738307386636734,
      "epoch": 8.95,
      "learning_rate": 4.105744337128662e-05,
      "loss": 0.0698,
      "step": 9425,
      "task_loss": 0.0339224673807621
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04669392481446266,
      "epoch": 8.95,
      "learning_rate": 4.104927514425312e-05,
      "loss": 0.0546,
      "step": 9426,
      "task_loss": 0.12560123205184937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04168114811182022,
      "epoch": 8.95,
      "learning_rate": 4.104110400179148e-05,
      "loss": 0.039,
      "step": 9427,
      "task_loss": 0.015049118548631668
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020040128380060196,
      "epoch": 8.95,
      "learning_rate": 4.103292994538605e-05,
      "loss": 0.0184,
      "step": 9428,
      "task_loss": 0.00409487821161747
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07207145541906357,
      "epoch": 8.95,
      "learning_rate": 4.102475297652168e-05,
      "loss": 0.0685,
      "step": 9429,
      "task_loss": 0.036523304879665375
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01679806597530842,
      "epoch": 8.96,
      "learning_rate": 4.1016573096683765e-05,
      "loss": 0.0215,
      "step": 9430,
      "task_loss": 0.0642915591597557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020311608910560608,
      "epoch": 8.96,
      "learning_rate": 4.10083903073582e-05,
      "loss": 0.0239,
      "step": 9431,
      "task_loss": 0.05604963004589081
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02877676486968994,
      "epoch": 8.96,
      "learning_rate": 4.1000204610031447e-05,
      "loss": 0.0267,
      "step": 9432,
      "task_loss": 0.007656911388039589
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024545563384890556,
      "epoch": 8.96,
      "learning_rate": 4.0992016006190456e-05,
      "loss": 0.0326,
      "step": 9433,
      "task_loss": 0.10507547855377197
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05073285847902298,
      "epoch": 8.96,
      "learning_rate": 4.0983824497322755e-05,
      "loss": 0.0484,
      "step": 9434,
      "task_loss": 0.027833662927150726
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027377085760235786,
      "epoch": 8.96,
      "learning_rate": 4.0975630084916344e-05,
      "loss": 0.0253,
      "step": 9435,
      "task_loss": 0.0062708742916584015
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015949569642543793,
      "epoch": 8.96,
      "learning_rate": 4.096743277045979e-05,
      "loss": 0.0235,
      "step": 9436,
      "task_loss": 0.09101761877536774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06154026836156845,
      "epoch": 8.96,
      "learning_rate": 4.0959232555442174e-05,
      "loss": 0.0634,
      "step": 9437,
      "task_loss": 0.07965725660324097
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.039994481950998306,
      "epoch": 8.96,
      "learning_rate": 4.0951029441353104e-05,
      "loss": 0.0373,
      "step": 9438,
      "task_loss": 0.012804200872778893
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07882995158433914,
      "epoch": 8.96,
      "learning_rate": 4.094282342968271e-05,
      "loss": 0.0807,
      "step": 9439,
      "task_loss": 0.09789763391017914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09340311586856842,
      "epoch": 8.96,
      "learning_rate": 4.093461452192167e-05,
      "loss": 0.0986,
      "step": 9440,
      "task_loss": 0.1451651006937027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0292808897793293,
      "epoch": 8.97,
      "learning_rate": 4.092640271956115e-05,
      "loss": 0.0349,
      "step": 9441,
      "task_loss": 0.08514495939016342
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1028076782822609,
      "epoch": 8.97,
      "learning_rate": 4.091818802409288e-05,
      "loss": 0.112,
      "step": 9442,
      "task_loss": 0.19489334523677826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10867805033922195,
      "epoch": 8.97,
      "learning_rate": 4.0909970437009096e-05,
      "loss": 0.1137,
      "step": 9443,
      "task_loss": 0.1586572676897049
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012916838750243187,
      "epoch": 8.97,
      "learning_rate": 4.0901749959802546e-05,
      "loss": 0.0168,
      "step": 9444,
      "task_loss": 0.05206574872136116
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1345238983631134,
      "epoch": 8.97,
      "learning_rate": 4.0893526593966535e-05,
      "loss": 0.1296,
      "step": 9445,
      "task_loss": 0.0848974958062172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07488267123699188,
      "epoch": 8.97,
      "learning_rate": 4.088530034099488e-05,
      "loss": 0.0872,
      "step": 9446,
      "task_loss": 0.1976650059223175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.041900884360075,
      "epoch": 8.97,
      "learning_rate": 4.087707120238191e-05,
      "loss": 0.0388,
      "step": 9447,
      "task_loss": 0.010816860944032669
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019380036741495132,
      "epoch": 8.97,
      "learning_rate": 4.0868839179622495e-05,
      "loss": 0.0188,
      "step": 9448,
      "task_loss": 0.013160983100533485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07051914930343628,
      "epoch": 8.97,
      "learning_rate": 4.086060427421202e-05,
      "loss": 0.0652,
      "step": 9449,
      "task_loss": 0.01760847680270672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0432322658598423,
      "epoch": 8.97,
      "learning_rate": 4.0852366487646384e-05,
      "loss": 0.0517,
      "step": 9450,
      "task_loss": 0.12742291390895844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027135606855154037,
      "epoch": 8.98,
      "learning_rate": 4.084412582142204e-05,
      "loss": 0.0364,
      "step": 9451,
      "task_loss": 0.1198822483420372
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.15044289827346802,
      "epoch": 8.98,
      "learning_rate": 4.083588227703593e-05,
      "loss": 0.1615,
      "step": 9452,
      "task_loss": 0.26102936267852783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0592607781291008,
      "epoch": 8.98,
      "learning_rate": 4.0827635855985534e-05,
      "loss": 0.0603,
      "step": 9453,
      "task_loss": 0.06970083713531494
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09907069802284241,
      "epoch": 8.98,
      "learning_rate": 4.081938655976886e-05,
      "loss": 0.1052,
      "step": 9454,
      "task_loss": 0.16077247262001038
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028371935710310936,
      "epoch": 8.98,
      "learning_rate": 4.0811134389884433e-05,
      "loss": 0.0264,
      "step": 9455,
      "task_loss": 0.008428094908595085
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0345245860517025,
      "epoch": 8.98,
      "learning_rate": 4.08028793478313e-05,
      "loss": 0.0411,
      "step": 9456,
      "task_loss": 0.09978184103965759
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.049868617206811905,
      "epoch": 8.98,
      "learning_rate": 4.0794621435109015e-05,
      "loss": 0.0674,
      "step": 9457,
      "task_loss": 0.22498998045921326
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02840796858072281,
      "epoch": 8.98,
      "learning_rate": 4.0786360653217684e-05,
      "loss": 0.0265,
      "step": 9458,
      "task_loss": 0.009237809106707573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03639814257621765,
      "epoch": 8.98,
      "learning_rate": 4.0778097003657915e-05,
      "loss": 0.0344,
      "step": 9459,
      "task_loss": 0.01689404621720314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020641740411520004,
      "epoch": 8.98,
      "learning_rate": 4.0769830487930835e-05,
      "loss": 0.0251,
      "step": 9460,
      "task_loss": 0.0654524564743042
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05284133553504944,
      "epoch": 8.98,
      "learning_rate": 4.07615611075381e-05,
      "loss": 0.0577,
      "step": 9461,
      "task_loss": 0.10118487477302551
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08153997361660004,
      "epoch": 8.99,
      "learning_rate": 4.075328886398188e-05,
      "loss": 0.0832,
      "step": 9462,
      "task_loss": 0.09851934015750885
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024140600115060806,
      "epoch": 8.99,
      "learning_rate": 4.074501375876487e-05,
      "loss": 0.0225,
      "step": 9463,
      "task_loss": 0.008064748719334602
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04541175067424774,
      "epoch": 8.99,
      "learning_rate": 4.073673579339028e-05,
      "loss": 0.0472,
      "step": 9464,
      "task_loss": 0.06317553669214249
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10922182351350784,
      "epoch": 8.99,
      "learning_rate": 4.0728454969361854e-05,
      "loss": 0.104,
      "step": 9465,
      "task_loss": 0.05705301836133003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04132990166544914,
      "epoch": 8.99,
      "learning_rate": 4.0720171288183815e-05,
      "loss": 0.0419,
      "step": 9466,
      "task_loss": 0.04671701043844223
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0743739902973175,
      "epoch": 8.99,
      "learning_rate": 4.0711884751360964e-05,
      "loss": 0.0726,
      "step": 9467,
      "task_loss": 0.05616597831249237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05336568504571915,
      "epoch": 8.99,
      "learning_rate": 4.070359536039858e-05,
      "loss": 0.0536,
      "step": 9468,
      "task_loss": 0.055919770151376724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12312227487564087,
      "epoch": 8.99,
      "learning_rate": 4.069530311680247e-05,
      "loss": 0.1289,
      "step": 9469,
      "task_loss": 0.180916890501976
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0300409235060215,
      "epoch": 8.99,
      "learning_rate": 4.068700802207895e-05,
      "loss": 0.0301,
      "step": 9470,
      "task_loss": 0.030945636332035065
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.059253282845020294,
      "epoch": 8.99,
      "learning_rate": 4.0678710077734885e-05,
      "loss": 0.0671,
      "step": 9471,
      "task_loss": 0.13786810636520386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02918911725282669,
      "epoch": 9.0,
      "learning_rate": 4.0670409285277614e-05,
      "loss": 0.0329,
      "step": 9472,
      "task_loss": 0.06589201092720032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1608899086713791,
      "epoch": 9.0,
      "learning_rate": 4.0662105646215034e-05,
      "loss": 0.1544,
      "step": 9473,
      "task_loss": 0.0955737829208374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.045890383422374725,
      "epoch": 9.0,
      "learning_rate": 4.065379916205554e-05,
      "loss": 0.0452,
      "step": 9474,
      "task_loss": 0.03897964581847191
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0502871535718441,
      "epoch": 9.0,
      "learning_rate": 4.0645489834308024e-05,
      "loss": 0.0519,
      "step": 9475,
      "task_loss": 0.06652585417032242
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12040196359157562,
      "epoch": 9.0,
      "learning_rate": 4.063717766448194e-05,
      "loss": 0.1166,
      "step": 9476,
      "task_loss": 0.08265364170074463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09221913665533066,
      "epoch": 9.0,
      "learning_rate": 4.062886265408722e-05,
      "loss": 0.0868,
      "step": 9477,
      "task_loss": 0.037736404687166214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08050169050693512,
      "epoch": 9.0,
      "learning_rate": 4.062054480463433e-05,
      "loss": 0.0768,
      "step": 9478,
      "task_loss": 0.043371133506298065
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02520603872835636,
      "epoch": 9.0,
      "learning_rate": 4.0612224117634245e-05,
      "loss": 0.0235,
      "step": 9479,
      "task_loss": 0.00836150161921978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018169641494750977,
      "epoch": 9.0,
      "learning_rate": 4.060390059459846e-05,
      "loss": 0.0285,
      "step": 9480,
      "task_loss": 0.1218915730714798
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09185026586055756,
      "epoch": 9.0,
      "learning_rate": 4.059557423703899e-05,
      "loss": 0.0925,
      "step": 9481,
      "task_loss": 0.09815002977848053
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016890790313482285,
      "epoch": 9.0,
      "learning_rate": 4.058724504646834e-05,
      "loss": 0.024,
      "step": 9482,
      "task_loss": 0.08820458501577377
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01857912167906761,
      "epoch": 9.01,
      "learning_rate": 4.0578913024399564e-05,
      "loss": 0.0222,
      "step": 9483,
      "task_loss": 0.05494304373860359
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0701369047164917,
      "epoch": 9.01,
      "learning_rate": 4.057057817234621e-05,
      "loss": 0.0669,
      "step": 9484,
      "task_loss": 0.03729560971260071
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017086252570152283,
      "epoch": 9.01,
      "learning_rate": 4.0562240491822334e-05,
      "loss": 0.0283,
      "step": 9485,
      "task_loss": 0.1295742690563202
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0375320240855217,
      "epoch": 9.01,
      "learning_rate": 4.055389998434253e-05,
      "loss": 0.0376,
      "step": 9486,
      "task_loss": 0.03832355886697769
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022690830752253532,
      "epoch": 9.01,
      "learning_rate": 4.054555665142189e-05,
      "loss": 0.0255,
      "step": 9487,
      "task_loss": 0.05128246918320656
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011978648602962494,
      "epoch": 9.01,
      "learning_rate": 4.053721049457601e-05,
      "loss": 0.0112,
      "step": 9488,
      "task_loss": 0.004542630165815353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014200868085026741,
      "epoch": 9.01,
      "learning_rate": 4.052886151532101e-05,
      "loss": 0.0218,
      "step": 9489,
      "task_loss": 0.09042149782180786
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.049319952726364136,
      "epoch": 9.01,
      "learning_rate": 4.0520509715173544e-05,
      "loss": 0.046,
      "step": 9490,
      "task_loss": 0.016364455223083496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024487420916557312,
      "epoch": 9.01,
      "learning_rate": 4.051215509565073e-05,
      "loss": 0.0227,
      "step": 9491,
      "task_loss": 0.006308834999799728
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018296554684638977,
      "epoch": 9.01,
      "learning_rate": 4.050379765827024e-05,
      "loss": 0.0169,
      "step": 9492,
      "task_loss": 0.003857152536511421
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04161164537072182,
      "epoch": 9.02,
      "learning_rate": 4.0495437404550233e-05,
      "loss": 0.0451,
      "step": 9493,
      "task_loss": 0.07690572738647461
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022137103602290154,
      "epoch": 9.02,
      "learning_rate": 4.04870743360094e-05,
      "loss": 0.0273,
      "step": 9494,
      "task_loss": 0.07371386885643005
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01800430938601494,
      "epoch": 9.02,
      "learning_rate": 4.047870845416693e-05,
      "loss": 0.017,
      "step": 9495,
      "task_loss": 0.00811527669429779
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024769477546215057,
      "epoch": 9.02,
      "learning_rate": 4.0470339760542506e-05,
      "loss": 0.0336,
      "step": 9496,
      "task_loss": 0.11265938729047775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05644622817635536,
      "epoch": 9.02,
      "learning_rate": 4.0461968256656376e-05,
      "loss": 0.055,
      "step": 9497,
      "task_loss": 0.042227305471897125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010226700454950333,
      "epoch": 9.02,
      "learning_rate": 4.045359394402925e-05,
      "loss": 0.0097,
      "step": 9498,
      "task_loss": 0.005248824134469032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020847581326961517,
      "epoch": 9.02,
      "learning_rate": 4.0445216824182344e-05,
      "loss": 0.0196,
      "step": 9499,
      "task_loss": 0.008282596245408058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11503365635871887,
      "epoch": 9.02,
      "learning_rate": 4.043683689863742e-05,
      "loss": 0.1119,
      "step": 9500,
      "task_loss": 0.08395393192768097
    },
    {
      "epoch": 9.02,
      "eval_accuracy": 0.8910550458715596,
      "eval_loss": 0.4567631483078003,
      "eval_runtime": 18.0265,
      "eval_samples_per_second": 48.373,
      "eval_steps_per_second": 6.047,
      "step": 9500
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04205450415611267,
      "epoch": 9.02,
      "learning_rate": 4.042845416891673e-05,
      "loss": 0.0486,
      "step": 9501,
      "task_loss": 0.10712815821170807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06236998736858368,
      "epoch": 9.02,
      "learning_rate": 4.042006863654303e-05,
      "loss": 0.0662,
      "step": 9502,
      "task_loss": 0.10055098682641983
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019925886765122414,
      "epoch": 9.02,
      "learning_rate": 4.041168030303961e-05,
      "loss": 0.0187,
      "step": 9503,
      "task_loss": 0.008159628137946129
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06345760822296143,
      "epoch": 9.03,
      "learning_rate": 4.0403289169930235e-05,
      "loss": 0.0664,
      "step": 9504,
      "task_loss": 0.09281642735004425
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023364584892988205,
      "epoch": 9.03,
      "learning_rate": 4.03948952387392e-05,
      "loss": 0.0272,
      "step": 9505,
      "task_loss": 0.06141924858093262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02729855477809906,
      "epoch": 9.03,
      "learning_rate": 4.03864985109913e-05,
      "loss": 0.0263,
      "step": 9506,
      "task_loss": 0.016974736005067825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01486403401941061,
      "epoch": 9.03,
      "learning_rate": 4.0378098988211845e-05,
      "loss": 0.0137,
      "step": 9507,
      "task_loss": 0.0035839397460222244
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.031053446233272552,
      "epoch": 9.03,
      "learning_rate": 4.036969667192665e-05,
      "loss": 0.0616,
      "step": 9508,
      "task_loss": 0.3368293344974518
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07761267572641373,
      "epoch": 9.03,
      "learning_rate": 4.036129156366203e-05,
      "loss": 0.0834,
      "step": 9509,
      "task_loss": 0.13566073775291443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1153075098991394,
      "epoch": 9.03,
      "learning_rate": 4.0352883664944816e-05,
      "loss": 0.11,
      "step": 9510,
      "task_loss": 0.06233564764261246
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015643972903490067,
      "epoch": 9.03,
      "learning_rate": 4.034447297730234e-05,
      "loss": 0.0146,
      "step": 9511,
      "task_loss": 0.005513627082109451
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04660728946328163,
      "epoch": 9.03,
      "learning_rate": 4.033605950226246e-05,
      "loss": 0.0524,
      "step": 9512,
      "task_loss": 0.10417984426021576
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01021486520767212,
      "epoch": 9.03,
      "learning_rate": 4.03276432413535e-05,
      "loss": 0.0097,
      "step": 9513,
      "task_loss": 0.004828939214348793
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030726449564099312,
      "epoch": 9.04,
      "learning_rate": 4.0319224196104334e-05,
      "loss": 0.0285,
      "step": 9514,
      "task_loss": 0.00808459147810936
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0746147632598877,
      "epoch": 9.04,
      "learning_rate": 4.031080236804431e-05,
      "loss": 0.0731,
      "step": 9515,
      "task_loss": 0.05903128162026405
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06045572832226753,
      "epoch": 9.04,
      "learning_rate": 4.03023777587033e-05,
      "loss": 0.073,
      "step": 9516,
      "task_loss": 0.18590320646762848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015560157597064972,
      "epoch": 9.04,
      "learning_rate": 4.029395036961168e-05,
      "loss": 0.0148,
      "step": 9517,
      "task_loss": 0.007637334987521172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02795090712606907,
      "epoch": 9.04,
      "learning_rate": 4.028552020230031e-05,
      "loss": 0.0278,
      "step": 9518,
      "task_loss": 0.026432598009705544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016976818442344666,
      "epoch": 9.04,
      "learning_rate": 4.0277087258300575e-05,
      "loss": 0.0156,
      "step": 9519,
      "task_loss": 0.0029955413192510605
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010076291859149933,
      "epoch": 9.04,
      "learning_rate": 4.0268651539144374e-05,
      "loss": 0.0094,
      "step": 9520,
      "task_loss": 0.003336844965815544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04456236958503723,
      "epoch": 9.04,
      "learning_rate": 4.026021304636408e-05,
      "loss": 0.0541,
      "step": 9521,
      "task_loss": 0.13946330547332764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02801411598920822,
      "epoch": 9.04,
      "learning_rate": 4.0251771781492594e-05,
      "loss": 0.0277,
      "step": 9522,
      "task_loss": 0.02517678588628769
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02441570535302162,
      "epoch": 9.04,
      "learning_rate": 4.0243327746063315e-05,
      "loss": 0.0303,
      "step": 9523,
      "task_loss": 0.08282805979251862
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03433768451213837,
      "epoch": 9.04,
      "learning_rate": 4.0234880941610134e-05,
      "loss": 0.0363,
      "step": 9524,
      "task_loss": 0.05413848161697388
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016913043335080147,
      "epoch": 9.05,
      "learning_rate": 4.022643136966746e-05,
      "loss": 0.026,
      "step": 9525,
      "task_loss": 0.10786878317594528
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07082056999206543,
      "epoch": 9.05,
      "learning_rate": 4.021797903177019e-05,
      "loss": 0.0752,
      "step": 9526,
      "task_loss": 0.114901602268219
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011534811928868294,
      "epoch": 9.05,
      "learning_rate": 4.0209523929453744e-05,
      "loss": 0.0107,
      "step": 9527,
      "task_loss": 0.002887837588787079
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018662162125110626,
      "epoch": 9.05,
      "learning_rate": 4.0201066064254026e-05,
      "loss": 0.0173,
      "step": 9528,
      "task_loss": 0.004804331809282303
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1188516914844513,
      "epoch": 9.05,
      "learning_rate": 4.019260543770745e-05,
      "loss": 0.1143,
      "step": 9529,
      "task_loss": 0.07374230027198792
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022575611248612404,
      "epoch": 9.05,
      "learning_rate": 4.018414205135093e-05,
      "loss": 0.0265,
      "step": 9530,
      "task_loss": 0.06145612150430679
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012420643121004105,
      "epoch": 9.05,
      "learning_rate": 4.017567590672187e-05,
      "loss": 0.0117,
      "step": 9531,
      "task_loss": 0.005104459822177887
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.032812122255563736,
      "epoch": 9.05,
      "learning_rate": 4.01672070053582e-05,
      "loss": 0.0399,
      "step": 9532,
      "task_loss": 0.10327274352312088
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023092597723007202,
      "epoch": 9.05,
      "learning_rate": 4.015873534879833e-05,
      "loss": 0.0245,
      "step": 9533,
      "task_loss": 0.03673432022333145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.037473469972610474,
      "epoch": 9.05,
      "learning_rate": 4.015026093858119e-05,
      "loss": 0.0472,
      "step": 9534,
      "task_loss": 0.13475218415260315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05353183299303055,
      "epoch": 9.06,
      "learning_rate": 4.014178377624617e-05,
      "loss": 0.0514,
      "step": 9535,
      "task_loss": 0.032544951885938644
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02789144590497017,
      "epoch": 9.06,
      "learning_rate": 4.013330386333321e-05,
      "loss": 0.0303,
      "step": 9536,
      "task_loss": 0.052158139646053314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009651558473706245,
      "epoch": 9.06,
      "learning_rate": 4.012482120138272e-05,
      "loss": 0.0091,
      "step": 9537,
      "task_loss": 0.003987642005085945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0357995480298996,
      "epoch": 9.06,
      "learning_rate": 4.011633579193561e-05,
      "loss": 0.0406,
      "step": 9538,
      "task_loss": 0.08331019431352615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017858661711215973,
      "epoch": 9.06,
      "learning_rate": 4.010784763653331e-05,
      "loss": 0.0252,
      "step": 9539,
      "task_loss": 0.09111450612545013
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06131261587142944,
      "epoch": 9.06,
      "learning_rate": 4.0099356736717725e-05,
      "loss": 0.0596,
      "step": 9540,
      "task_loss": 0.04372568055987358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021268391981720924,
      "epoch": 9.06,
      "learning_rate": 4.0090863094031274e-05,
      "loss": 0.0294,
      "step": 9541,
      "task_loss": 0.10256145894527435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030995316803455353,
      "epoch": 9.06,
      "learning_rate": 4.008236671001686e-05,
      "loss": 0.0286,
      "step": 9542,
      "task_loss": 0.007312217727303505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016386138275265694,
      "epoch": 9.06,
      "learning_rate": 4.0073867586217895e-05,
      "loss": 0.0153,
      "step": 9543,
      "task_loss": 0.005437880754470825
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02716147154569626,
      "epoch": 9.06,
      "learning_rate": 4.006536572417828e-05,
      "loss": 0.0256,
      "step": 9544,
      "task_loss": 0.011989755555987358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03949110209941864,
      "epoch": 9.06,
      "learning_rate": 4.0056861125442435e-05,
      "loss": 0.0435,
      "step": 9545,
      "task_loss": 0.07948499917984009
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.060796599835157394,
      "epoch": 9.07,
      "learning_rate": 4.004835379155525e-05,
      "loss": 0.0645,
      "step": 9546,
      "task_loss": 0.09797711670398712
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.00800246186554432,
      "epoch": 9.07,
      "learning_rate": 4.003984372406212e-05,
      "loss": 0.0078,
      "step": 9547,
      "task_loss": 0.0055456701666116714
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01699589565396309,
      "epoch": 9.07,
      "learning_rate": 4.003133092450895e-05,
      "loss": 0.0158,
      "step": 9548,
      "task_loss": 0.005228973925113678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0232367105782032,
      "epoch": 9.07,
      "learning_rate": 4.002281539444213e-05,
      "loss": 0.03,
      "step": 9549,
      "task_loss": 0.09088429808616638
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027070309966802597,
      "epoch": 9.07,
      "learning_rate": 4.001429713540853e-05,
      "loss": 0.0249,
      "step": 9550,
      "task_loss": 0.005381651222705841
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030805271118879318,
      "epoch": 9.07,
      "learning_rate": 4.000577614895555e-05,
      "loss": 0.031,
      "step": 9551,
      "task_loss": 0.03288976103067398
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.15201908349990845,
      "epoch": 9.07,
      "learning_rate": 3.999725243663107e-05,
      "loss": 0.1577,
      "step": 9552,
      "task_loss": 0.2092512547969818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03263195604085922,
      "epoch": 9.07,
      "learning_rate": 3.9988725999983456e-05,
      "loss": 0.0373,
      "step": 9553,
      "task_loss": 0.07890980690717697
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13403475284576416,
      "epoch": 9.07,
      "learning_rate": 3.998019684056158e-05,
      "loss": 0.1317,
      "step": 9554,
      "task_loss": 0.11034642159938812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04724042862653732,
      "epoch": 9.07,
      "learning_rate": 3.99716649599148e-05,
      "loss": 0.0483,
      "step": 9555,
      "task_loss": 0.058092519640922546
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0314975380897522,
      "epoch": 9.08,
      "learning_rate": 3.996313035959297e-05,
      "loss": 0.037,
      "step": 9556,
      "task_loss": 0.0861077755689621
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01337359193712473,
      "epoch": 9.08,
      "learning_rate": 3.995459304114645e-05,
      "loss": 0.0126,
      "step": 9557,
      "task_loss": 0.005379321053624153
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02194923162460327,
      "epoch": 9.08,
      "learning_rate": 3.9946053006126086e-05,
      "loss": 0.0286,
      "step": 9558,
      "task_loss": 0.08856151252985
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021245867013931274,
      "epoch": 9.08,
      "learning_rate": 3.993751025608321e-05,
      "loss": 0.0271,
      "step": 9559,
      "task_loss": 0.0795937329530716
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025511398911476135,
      "epoch": 9.08,
      "learning_rate": 3.9928964792569655e-05,
      "loss": 0.0235,
      "step": 9560,
      "task_loss": 0.005805861204862595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018126703798770905,
      "epoch": 9.08,
      "learning_rate": 3.9920416617137745e-05,
      "loss": 0.017,
      "step": 9561,
      "task_loss": 0.007210768759250641
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018573222681879997,
      "epoch": 9.08,
      "learning_rate": 3.9911865731340306e-05,
      "loss": 0.0233,
      "step": 9562,
      "task_loss": 0.06605537980794907
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023824790492653847,
      "epoch": 9.08,
      "learning_rate": 3.9903312136730634e-05,
      "loss": 0.0321,
      "step": 9563,
      "task_loss": 0.10653051733970642
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10854936391115189,
      "epoch": 9.08,
      "learning_rate": 3.989475583486254e-05,
      "loss": 0.105,
      "step": 9564,
      "task_loss": 0.07346247136592865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019099000841379166,
      "epoch": 9.08,
      "learning_rate": 3.988619682729032e-05,
      "loss": 0.0268,
      "step": 9565,
      "task_loss": 0.09585027396678925
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021908055990934372,
      "epoch": 9.08,
      "learning_rate": 3.987763511556874e-05,
      "loss": 0.0296,
      "step": 9566,
      "task_loss": 0.09896925091743469
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013894637115299702,
      "epoch": 9.09,
      "learning_rate": 3.98690707012531e-05,
      "loss": 0.0128,
      "step": 9567,
      "task_loss": 0.0032347403466701508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03824283555150032,
      "epoch": 9.09,
      "learning_rate": 3.986050358589916e-05,
      "loss": 0.0488,
      "step": 9568,
      "task_loss": 0.14340779185295105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03835334628820419,
      "epoch": 9.09,
      "learning_rate": 3.9851933771063166e-05,
      "loss": 0.0349,
      "step": 9569,
      "task_loss": 0.003623614087700844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03116435743868351,
      "epoch": 9.09,
      "learning_rate": 3.9843361258301876e-05,
      "loss": 0.0385,
      "step": 9570,
      "task_loss": 0.10502012073993683
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04906230419874191,
      "epoch": 9.09,
      "learning_rate": 3.983478604917253e-05,
      "loss": 0.0574,
      "step": 9571,
      "task_loss": 0.13289552927017212
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04502825438976288,
      "epoch": 9.09,
      "learning_rate": 3.9826208145232855e-05,
      "loss": 0.0679,
      "step": 9572,
      "task_loss": 0.273649126291275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.036221615970134735,
      "epoch": 9.09,
      "learning_rate": 3.981762754804107e-05,
      "loss": 0.0414,
      "step": 9573,
      "task_loss": 0.08760883659124374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05786776542663574,
      "epoch": 9.09,
      "learning_rate": 3.980904425915586e-05,
      "loss": 0.0567,
      "step": 9574,
      "task_loss": 0.04585399478673935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0210392028093338,
      "epoch": 9.09,
      "learning_rate": 3.9800458280136453e-05,
      "loss": 0.0198,
      "step": 9575,
      "task_loss": 0.008328957483172417
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.047202907502651215,
      "epoch": 9.09,
      "learning_rate": 3.979186961254252e-05,
      "loss": 0.0497,
      "step": 9576,
      "task_loss": 0.07255647331476212
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02497425302863121,
      "epoch": 9.09,
      "learning_rate": 3.9783278257934233e-05,
      "loss": 0.0233,
      "step": 9577,
      "task_loss": 0.00862952508032322
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.031044363975524902,
      "epoch": 9.1,
      "learning_rate": 3.977468421787225e-05,
      "loss": 0.0361,
      "step": 9578,
      "task_loss": 0.08184700459241867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023872077465057373,
      "epoch": 9.1,
      "learning_rate": 3.976608749391773e-05,
      "loss": 0.0291,
      "step": 9579,
      "task_loss": 0.0763431191444397
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021440522745251656,
      "epoch": 9.1,
      "learning_rate": 3.975748808763229e-05,
      "loss": 0.0202,
      "step": 9580,
      "task_loss": 0.009024685248732567
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017536599189043045,
      "epoch": 9.1,
      "learning_rate": 3.974888600057808e-05,
      "loss": 0.0187,
      "step": 9581,
      "task_loss": 0.029022112488746643
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03331207111477852,
      "epoch": 9.1,
      "learning_rate": 3.974028123431769e-05,
      "loss": 0.0486,
      "step": 9582,
      "task_loss": 0.1862955540418625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01485717948526144,
      "epoch": 9.1,
      "learning_rate": 3.973167379041421e-05,
      "loss": 0.0142,
      "step": 9583,
      "task_loss": 0.0080320555716753
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03911569342017174,
      "epoch": 9.1,
      "learning_rate": 3.972306367043126e-05,
      "loss": 0.043,
      "step": 9584,
      "task_loss": 0.07806295156478882
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1487078070640564,
      "epoch": 9.1,
      "learning_rate": 3.971445087593288e-05,
      "loss": 0.1381,
      "step": 9585,
      "task_loss": 0.04280809685587883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01248634047806263,
      "epoch": 9.1,
      "learning_rate": 3.970583540848363e-05,
      "loss": 0.0155,
      "step": 9586,
      "task_loss": 0.04234500974416733
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015881624072790146,
      "epoch": 9.1,
      "learning_rate": 3.969721726964856e-05,
      "loss": 0.0236,
      "step": 9587,
      "task_loss": 0.09326736629009247
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028029238805174828,
      "epoch": 9.11,
      "learning_rate": 3.9688596460993176e-05,
      "loss": 0.0332,
      "step": 9588,
      "task_loss": 0.07947677373886108
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018526611849665642,
      "epoch": 9.11,
      "learning_rate": 3.967997298408352e-05,
      "loss": 0.0217,
      "step": 9589,
      "task_loss": 0.04995952919125557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02319607511162758,
      "epoch": 9.11,
      "learning_rate": 3.967134684048607e-05,
      "loss": 0.0304,
      "step": 9590,
      "task_loss": 0.09573376178741455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03988368809223175,
      "epoch": 9.11,
      "learning_rate": 3.96627180317678e-05,
      "loss": 0.0397,
      "step": 9591,
      "task_loss": 0.03814127668738365
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030341364443302155,
      "epoch": 9.11,
      "learning_rate": 3.965408655949619e-05,
      "loss": 0.0277,
      "step": 9592,
      "task_loss": 0.004290319979190826
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018847042694687843,
      "epoch": 9.11,
      "learning_rate": 3.964545242523917e-05,
      "loss": 0.0174,
      "step": 9593,
      "task_loss": 0.004647746682167053
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05705046281218529,
      "epoch": 9.11,
      "learning_rate": 3.9636815630565194e-05,
      "loss": 0.055,
      "step": 9594,
      "task_loss": 0.03697717562317848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02136809565126896,
      "epoch": 9.11,
      "learning_rate": 3.962817617704317e-05,
      "loss": 0.0265,
      "step": 9595,
      "task_loss": 0.07274884730577469
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.2957301437854767,
      "epoch": 9.11,
      "learning_rate": 3.9619534066242485e-05,
      "loss": 0.281,
      "step": 9596,
      "task_loss": 0.14882266521453857
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018483705818653107,
      "epoch": 9.11,
      "learning_rate": 3.961088929973303e-05,
      "loss": 0.0266,
      "step": 9597,
      "task_loss": 0.09919779002666473
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.031163640320301056,
      "epoch": 9.11,
      "learning_rate": 3.960224187908518e-05,
      "loss": 0.0309,
      "step": 9598,
      "task_loss": 0.028584491461515427
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0134575255215168,
      "epoch": 9.12,
      "learning_rate": 3.959359180586975e-05,
      "loss": 0.0128,
      "step": 9599,
      "task_loss": 0.006391104310750961
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020069237798452377,
      "epoch": 9.12,
      "learning_rate": 3.958493908165809e-05,
      "loss": 0.0185,
      "step": 9600,
      "task_loss": 0.004250597208738327
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04818892478942871,
      "epoch": 9.12,
      "learning_rate": 3.9576283708022e-05,
      "loss": 0.0446,
      "step": 9601,
      "task_loss": 0.012091221287846565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06722578406333923,
      "epoch": 9.12,
      "learning_rate": 3.956762568653378e-05,
      "loss": 0.0683,
      "step": 9602,
      "task_loss": 0.07760586589574814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021520771086215973,
      "epoch": 9.12,
      "learning_rate": 3.95589650187662e-05,
      "loss": 0.0263,
      "step": 9603,
      "task_loss": 0.06911582499742508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07241514325141907,
      "epoch": 9.12,
      "learning_rate": 3.95503017062925e-05,
      "loss": 0.0696,
      "step": 9604,
      "task_loss": 0.04403474181890488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0638948380947113,
      "epoch": 9.12,
      "learning_rate": 3.954163575068643e-05,
      "loss": 0.0619,
      "step": 9605,
      "task_loss": 0.04400301352143288
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014446464367210865,
      "epoch": 9.12,
      "learning_rate": 3.953296715352218e-05,
      "loss": 0.0136,
      "step": 9606,
      "task_loss": 0.0059725940227508545
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016270997002720833,
      "epoch": 9.12,
      "learning_rate": 3.952429591637446e-05,
      "loss": 0.0242,
      "step": 9607,
      "task_loss": 0.095741406083107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025010449811816216,
      "epoch": 9.12,
      "learning_rate": 3.951562204081845e-05,
      "loss": 0.0283,
      "step": 9608,
      "task_loss": 0.057582780718803406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07090044021606445,
      "epoch": 9.13,
      "learning_rate": 3.950694552842977e-05,
      "loss": 0.0913,
      "step": 9609,
      "task_loss": 0.2748509645462036
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022294355556368828,
      "epoch": 9.13,
      "learning_rate": 3.949826638078457e-05,
      "loss": 0.0367,
      "step": 9610,
      "task_loss": 0.16597087681293488
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02015480399131775,
      "epoch": 9.13,
      "learning_rate": 3.948958459945946e-05,
      "loss": 0.0314,
      "step": 9611,
      "task_loss": 0.13252711296081543
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06596086174249649,
      "epoch": 9.13,
      "learning_rate": 3.948090018603153e-05,
      "loss": 0.0796,
      "step": 9612,
      "task_loss": 0.20231035351753235
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018903814256191254,
      "epoch": 9.13,
      "learning_rate": 3.947221314207834e-05,
      "loss": 0.0329,
      "step": 9613,
      "task_loss": 0.15914104878902435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03513910248875618,
      "epoch": 9.13,
      "learning_rate": 3.9463523469177935e-05,
      "loss": 0.0369,
      "step": 9614,
      "task_loss": 0.05302010476589203
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04065433889627457,
      "epoch": 9.13,
      "learning_rate": 3.9454831168908824e-05,
      "loss": 0.0387,
      "step": 9615,
      "task_loss": 0.021478936076164246
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02802848257124424,
      "epoch": 9.13,
      "learning_rate": 3.9446136242850025e-05,
      "loss": 0.0257,
      "step": 9616,
      "task_loss": 0.004691721871495247
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09828861057758331,
      "epoch": 9.13,
      "learning_rate": 3.9437438692581e-05,
      "loss": 0.0998,
      "step": 9617,
      "task_loss": 0.11373132467269897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05762851983308792,
      "epoch": 9.13,
      "learning_rate": 3.9428738519681704e-05,
      "loss": 0.0569,
      "step": 9618,
      "task_loss": 0.04999490827322006
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03731340914964676,
      "epoch": 9.13,
      "learning_rate": 3.942003572573257e-05,
      "loss": 0.037,
      "step": 9619,
      "task_loss": 0.0341169647872448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023080751299858093,
      "epoch": 9.14,
      "learning_rate": 3.94113303123145e-05,
      "loss": 0.0346,
      "step": 9620,
      "task_loss": 0.13836701214313507
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014194730669260025,
      "epoch": 9.14,
      "learning_rate": 3.9402622281008874e-05,
      "loss": 0.0199,
      "step": 9621,
      "task_loss": 0.07124481350183487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05054982379078865,
      "epoch": 9.14,
      "learning_rate": 3.939391163339754e-05,
      "loss": 0.0559,
      "step": 9622,
      "task_loss": 0.10361889749765396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014216311275959015,
      "epoch": 9.14,
      "learning_rate": 3.9385198371062845e-05,
      "loss": 0.0241,
      "step": 9623,
      "task_loss": 0.11282174289226532
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014683485962450504,
      "epoch": 9.14,
      "learning_rate": 3.937648249558758e-05,
      "loss": 0.0136,
      "step": 9624,
      "task_loss": 0.0039719510823488235
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016595659777522087,
      "epoch": 9.14,
      "learning_rate": 3.9367764008555034e-05,
      "loss": 0.0156,
      "step": 9625,
      "task_loss": 0.006236037239432335
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015016734600067139,
      "epoch": 9.14,
      "learning_rate": 3.9359042911548955e-05,
      "loss": 0.014,
      "step": 9626,
      "task_loss": 0.005279116332530975
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012623798102140427,
      "epoch": 9.14,
      "learning_rate": 3.935031920615358e-05,
      "loss": 0.016,
      "step": 9627,
      "task_loss": 0.04630003124475479
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05201908200979233,
      "epoch": 9.14,
      "learning_rate": 3.934159289395361e-05,
      "loss": 0.0515,
      "step": 9628,
      "task_loss": 0.04690450802445412
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02428932674229145,
      "epoch": 9.14,
      "learning_rate": 3.9332863976534225e-05,
      "loss": 0.0316,
      "step": 9629,
      "task_loss": 0.0969204306602478
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03481846675276756,
      "epoch": 9.15,
      "learning_rate": 3.9324132455481064e-05,
      "loss": 0.0322,
      "step": 9630,
      "task_loss": 0.008588599041104317
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011342018842697144,
      "epoch": 9.15,
      "learning_rate": 3.931539833238026e-05,
      "loss": 0.0157,
      "step": 9631,
      "task_loss": 0.05512590333819389
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04356681555509567,
      "epoch": 9.15,
      "learning_rate": 3.930666160881841e-05,
      "loss": 0.0502,
      "step": 9632,
      "task_loss": 0.10991165041923523
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03675740957260132,
      "epoch": 9.15,
      "learning_rate": 3.9297922286382573e-05,
      "loss": 0.0385,
      "step": 9633,
      "task_loss": 0.05395536124706268
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012938008643686771,
      "epoch": 9.15,
      "learning_rate": 3.928918036666029e-05,
      "loss": 0.0123,
      "step": 9634,
      "task_loss": 0.006980478763580322
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.038404569029808044,
      "epoch": 9.15,
      "learning_rate": 3.928043585123957e-05,
      "loss": 0.0375,
      "step": 9635,
      "task_loss": 0.029215719550848007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04972174018621445,
      "epoch": 9.15,
      "learning_rate": 3.927168874170891e-05,
      "loss": 0.0643,
      "step": 9636,
      "task_loss": 0.19559511542320251
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0464179627597332,
      "epoch": 9.15,
      "learning_rate": 3.926293903965726e-05,
      "loss": 0.0465,
      "step": 9637,
      "task_loss": 0.04701643064618111
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06579018384218216,
      "epoch": 9.15,
      "learning_rate": 3.925418674667405e-05,
      "loss": 0.0745,
      "step": 9638,
      "task_loss": 0.1530209332704544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04052596539258957,
      "epoch": 9.15,
      "learning_rate": 3.924543186434915e-05,
      "loss": 0.0394,
      "step": 9639,
      "task_loss": 0.029542801901698112
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06306673586368561,
      "epoch": 9.15,
      "learning_rate": 3.923667439427295e-05,
      "loss": 0.066,
      "step": 9640,
      "task_loss": 0.09248776733875275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04041620343923569,
      "epoch": 9.16,
      "learning_rate": 3.922791433803629e-05,
      "loss": 0.0383,
      "step": 9641,
      "task_loss": 0.019168421626091003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015475798398256302,
      "epoch": 9.16,
      "learning_rate": 3.921915169723046e-05,
      "loss": 0.0145,
      "step": 9642,
      "task_loss": 0.005340602248907089
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12808610498905182,
      "epoch": 9.16,
      "learning_rate": 3.921038647344725e-05,
      "loss": 0.1262,
      "step": 9643,
      "task_loss": 0.10924281924962997
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04638529568910599,
      "epoch": 9.16,
      "learning_rate": 3.920161866827889e-05,
      "loss": 0.0483,
      "step": 9644,
      "task_loss": 0.06601787358522415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13820701837539673,
      "epoch": 9.16,
      "learning_rate": 3.9192848283318114e-05,
      "loss": 0.1419,
      "step": 9645,
      "task_loss": 0.17469622194766998
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0257248692214489,
      "epoch": 9.16,
      "learning_rate": 3.91840753201581e-05,
      "loss": 0.0321,
      "step": 9646,
      "task_loss": 0.08985202014446259
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.057688090950250626,
      "epoch": 9.16,
      "learning_rate": 3.917529978039247e-05,
      "loss": 0.0599,
      "step": 9647,
      "task_loss": 0.079569973051548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02995416149497032,
      "epoch": 9.16,
      "learning_rate": 3.9166521665615386e-05,
      "loss": 0.0362,
      "step": 9648,
      "task_loss": 0.09234865754842758
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014706909656524658,
      "epoch": 9.16,
      "learning_rate": 3.91577409774214e-05,
      "loss": 0.0136,
      "step": 9649,
      "task_loss": 0.0036916378885507584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04012545198202133,
      "epoch": 9.16,
      "learning_rate": 3.9148957717405596e-05,
      "loss": 0.0443,
      "step": 9650,
      "task_loss": 0.08144651353359222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03406708687543869,
      "epoch": 9.17,
      "learning_rate": 3.914017188716347e-05,
      "loss": 0.0402,
      "step": 9651,
      "task_loss": 0.09540209174156189
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022918783128261566,
      "epoch": 9.17,
      "learning_rate": 3.913138348829102e-05,
      "loss": 0.0348,
      "step": 9652,
      "task_loss": 0.14145193994045258
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013142495416104794,
      "epoch": 9.17,
      "learning_rate": 3.91225925223847e-05,
      "loss": 0.0121,
      "step": 9653,
      "task_loss": 0.0028307754546403885
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.036516182124614716,
      "epoch": 9.17,
      "learning_rate": 3.911379899104144e-05,
      "loss": 0.0346,
      "step": 9654,
      "task_loss": 0.016904115676879883
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04557647556066513,
      "epoch": 9.17,
      "learning_rate": 3.910500289585862e-05,
      "loss": 0.045,
      "step": 9655,
      "task_loss": 0.04023095220327377
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01984507404267788,
      "epoch": 9.17,
      "learning_rate": 3.90962042384341e-05,
      "loss": 0.0186,
      "step": 9656,
      "task_loss": 0.0074311550706624985
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0343051478266716,
      "epoch": 9.17,
      "learning_rate": 3.908740302036618e-05,
      "loss": 0.0397,
      "step": 9657,
      "task_loss": 0.08872069418430328
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024765335023403168,
      "epoch": 9.17,
      "learning_rate": 3.907859924325366e-05,
      "loss": 0.0381,
      "step": 9658,
      "task_loss": 0.15835905075073242
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12249832600355148,
      "epoch": 9.17,
      "learning_rate": 3.906979290869578e-05,
      "loss": 0.1308,
      "step": 9659,
      "task_loss": 0.20595303177833557
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012475738301873207,
      "epoch": 9.17,
      "learning_rate": 3.9060984018292267e-05,
      "loss": 0.0118,
      "step": 9660,
      "task_loss": 0.0056333523243665695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023209767416119576,
      "epoch": 9.17,
      "learning_rate": 3.905217257364328e-05,
      "loss": 0.0386,
      "step": 9661,
      "task_loss": 0.17690522968769073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04662572965025902,
      "epoch": 9.18,
      "learning_rate": 3.904335857634948e-05,
      "loss": 0.0445,
      "step": 9662,
      "task_loss": 0.0256513562053442
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04135457053780556,
      "epoch": 9.18,
      "learning_rate": 3.9034542028011944e-05,
      "loss": 0.0443,
      "step": 9663,
      "task_loss": 0.07109534740447998
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.036145225167274475,
      "epoch": 9.18,
      "learning_rate": 3.902572293023227e-05,
      "loss": 0.0458,
      "step": 9664,
      "task_loss": 0.13244393467903137
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03712350130081177,
      "epoch": 9.18,
      "learning_rate": 3.9016901284612474e-05,
      "loss": 0.0569,
      "step": 9665,
      "task_loss": 0.23489037156105042
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011418421752750874,
      "epoch": 9.18,
      "learning_rate": 3.9008077092755055e-05,
      "loss": 0.0162,
      "step": 9666,
      "task_loss": 0.058784160763025284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08879618346691132,
      "epoch": 9.18,
      "learning_rate": 3.899925035626296e-05,
      "loss": 0.0833,
      "step": 9667,
      "task_loss": 0.03357243910431862
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10656376928091049,
      "epoch": 9.18,
      "learning_rate": 3.899042107673962e-05,
      "loss": 0.1155,
      "step": 9668,
      "task_loss": 0.19606941938400269
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02124224603176117,
      "epoch": 9.18,
      "learning_rate": 3.898158925578893e-05,
      "loss": 0.0316,
      "step": 9669,
      "task_loss": 0.12506425380706787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023364577442407608,
      "epoch": 9.18,
      "learning_rate": 3.89727548950152e-05,
      "loss": 0.0221,
      "step": 9670,
      "task_loss": 0.01049577072262764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.054638948291540146,
      "epoch": 9.18,
      "learning_rate": 3.8963917996023245e-05,
      "loss": 0.0538,
      "step": 9671,
      "task_loss": 0.04668677970767021
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03254498541355133,
      "epoch": 9.19,
      "learning_rate": 3.8955078560418345e-05,
      "loss": 0.0449,
      "step": 9672,
      "task_loss": 0.15559810400009155
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020883910357952118,
      "epoch": 9.19,
      "learning_rate": 3.894623658980622e-05,
      "loss": 0.0193,
      "step": 9673,
      "task_loss": 0.005325049161911011
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1067768931388855,
      "epoch": 9.19,
      "learning_rate": 3.8937392085793036e-05,
      "loss": 0.1221,
      "step": 9674,
      "task_loss": 0.2597641348838806
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018741261214017868,
      "epoch": 9.19,
      "learning_rate": 3.892854504998546e-05,
      "loss": 0.029,
      "step": 9675,
      "task_loss": 0.12132866680622101
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011632721871137619,
      "epoch": 9.19,
      "learning_rate": 3.891969548399061e-05,
      "loss": 0.0203,
      "step": 9676,
      "task_loss": 0.09824814647436142
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017414584755897522,
      "epoch": 9.19,
      "learning_rate": 3.891084338941603e-05,
      "loss": 0.0162,
      "step": 9677,
      "task_loss": 0.00522034615278244
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012747041881084442,
      "epoch": 9.19,
      "learning_rate": 3.8901988767869744e-05,
      "loss": 0.0118,
      "step": 9678,
      "task_loss": 0.0028500892221927643
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.055840395390987396,
      "epoch": 9.19,
      "learning_rate": 3.8893131620960254e-05,
      "loss": 0.057,
      "step": 9679,
      "task_loss": 0.06704328209161758
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03462572395801544,
      "epoch": 9.19,
      "learning_rate": 3.88842719502965e-05,
      "loss": 0.0367,
      "step": 9680,
      "task_loss": 0.05576511472463608
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.050681374967098236,
      "epoch": 9.19,
      "learning_rate": 3.887540975748787e-05,
      "loss": 0.0543,
      "step": 9681,
      "task_loss": 0.08639328181743622
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029797300696372986,
      "epoch": 9.19,
      "learning_rate": 3.8866545044144234e-05,
      "loss": 0.0337,
      "step": 9682,
      "task_loss": 0.06878768652677536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014516827650368214,
      "epoch": 9.2,
      "learning_rate": 3.885767781187591e-05,
      "loss": 0.0236,
      "step": 9683,
      "task_loss": 0.10563208907842636
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013084099628031254,
      "epoch": 9.2,
      "learning_rate": 3.884880806229367e-05,
      "loss": 0.0139,
      "step": 9684,
      "task_loss": 0.020864391699433327
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022157765924930573,
      "epoch": 9.2,
      "learning_rate": 3.883993579700875e-05,
      "loss": 0.0206,
      "step": 9685,
      "task_loss": 0.006276823580265045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.059577975422143936,
      "epoch": 9.2,
      "learning_rate": 3.883106101763285e-05,
      "loss": 0.0571,
      "step": 9686,
      "task_loss": 0.03433872014284134
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04622458666563034,
      "epoch": 9.2,
      "learning_rate": 3.882218372577809e-05,
      "loss": 0.06,
      "step": 9687,
      "task_loss": 0.18399886786937714
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030847573652863503,
      "epoch": 9.2,
      "learning_rate": 3.881330392305709e-05,
      "loss": 0.0407,
      "step": 9688,
      "task_loss": 0.1292884647846222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10208854079246521,
      "epoch": 9.2,
      "learning_rate": 3.8804421611082916e-05,
      "loss": 0.1025,
      "step": 9689,
      "task_loss": 0.10571222007274628
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07040687650442123,
      "epoch": 9.2,
      "learning_rate": 3.8795536791469066e-05,
      "loss": 0.0801,
      "step": 9690,
      "task_loss": 0.16708990931510925
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022444941103458405,
      "epoch": 9.2,
      "learning_rate": 3.8786649465829516e-05,
      "loss": 0.0288,
      "step": 9691,
      "task_loss": 0.08632338047027588
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014007965102791786,
      "epoch": 9.2,
      "learning_rate": 3.8777759635778696e-05,
      "loss": 0.0129,
      "step": 9692,
      "task_loss": 0.003078687936067581
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03719889000058174,
      "epoch": 9.21,
      "learning_rate": 3.876886730293149e-05,
      "loss": 0.0361,
      "step": 9693,
      "task_loss": 0.0266589242964983
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01961357146501541,
      "epoch": 9.21,
      "learning_rate": 3.8759972468903215e-05,
      "loss": 0.0255,
      "step": 9694,
      "task_loss": 0.07875269651412964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021031856536865234,
      "epoch": 9.21,
      "learning_rate": 3.875107513530968e-05,
      "loss": 0.0209,
      "step": 9695,
      "task_loss": 0.01938748173415661
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0264435987919569,
      "epoch": 9.21,
      "learning_rate": 3.874217530376711e-05,
      "loss": 0.0375,
      "step": 9696,
      "task_loss": 0.13744567334651947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07807254046201706,
      "epoch": 9.21,
      "learning_rate": 3.873327297589223e-05,
      "loss": 0.0742,
      "step": 9697,
      "task_loss": 0.03972852602601051
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024312328547239304,
      "epoch": 9.21,
      "learning_rate": 3.8724368153302166e-05,
      "loss": 0.0274,
      "step": 9698,
      "task_loss": 0.055370982736349106
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01522772666066885,
      "epoch": 9.21,
      "learning_rate": 3.871546083761453e-05,
      "loss": 0.0185,
      "step": 9699,
      "task_loss": 0.04807128384709358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10017004609107971,
      "epoch": 9.21,
      "learning_rate": 3.870655103044738e-05,
      "loss": 0.1018,
      "step": 9700,
      "task_loss": 0.11637624353170395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04961914196610451,
      "epoch": 9.21,
      "learning_rate": 3.8697638733419216e-05,
      "loss": 0.0533,
      "step": 9701,
      "task_loss": 0.086252860724926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06641680747270584,
      "epoch": 9.21,
      "learning_rate": 3.8688723948149014e-05,
      "loss": 0.0691,
      "step": 9702,
      "task_loss": 0.0931624248623848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025418289005756378,
      "epoch": 9.21,
      "learning_rate": 3.867980667625618e-05,
      "loss": 0.0309,
      "step": 9703,
      "task_loss": 0.07991056889295578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017918124794960022,
      "epoch": 9.22,
      "learning_rate": 3.867088691936058e-05,
      "loss": 0.0227,
      "step": 9704,
      "task_loss": 0.06607376039028168
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.051767975091934204,
      "epoch": 9.22,
      "learning_rate": 3.8661964679082535e-05,
      "loss": 0.0597,
      "step": 9705,
      "task_loss": 0.13116154074668884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030053768306970596,
      "epoch": 9.22,
      "learning_rate": 3.8653039957042806e-05,
      "loss": 0.0285,
      "step": 9706,
      "task_loss": 0.014347141608595848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010007740929722786,
      "epoch": 9.22,
      "learning_rate": 3.8644112754862614e-05,
      "loss": 0.0093,
      "step": 9707,
      "task_loss": 0.002756282687187195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03509853407740593,
      "epoch": 9.22,
      "learning_rate": 3.8635183074163636e-05,
      "loss": 0.038,
      "step": 9708,
      "task_loss": 0.06392055749893188
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02037603035569191,
      "epoch": 9.22,
      "learning_rate": 3.862625091656797e-05,
      "loss": 0.0211,
      "step": 9709,
      "task_loss": 0.02797846868634224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04154079779982567,
      "epoch": 9.22,
      "learning_rate": 3.861731628369822e-05,
      "loss": 0.0651,
      "step": 9710,
      "task_loss": 0.2771263122558594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028145533055067062,
      "epoch": 9.22,
      "learning_rate": 3.8608379177177375e-05,
      "loss": 0.042,
      "step": 9711,
      "task_loss": 0.1668614000082016
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018877774477005005,
      "epoch": 9.22,
      "learning_rate": 3.8599439598628916e-05,
      "loss": 0.0314,
      "step": 9712,
      "task_loss": 0.14435160160064697
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.128778338432312,
      "epoch": 9.22,
      "learning_rate": 3.8590497549676753e-05,
      "loss": 0.1268,
      "step": 9713,
      "task_loss": 0.1085672602057457
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02545616589486599,
      "epoch": 9.23,
      "learning_rate": 3.858155303194526e-05,
      "loss": 0.0276,
      "step": 9714,
      "task_loss": 0.0471004918217659
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025828856974840164,
      "epoch": 9.23,
      "learning_rate": 3.8572606047059254e-05,
      "loss": 0.0246,
      "step": 9715,
      "task_loss": 0.01345803216099739
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018282314762473106,
      "epoch": 9.23,
      "learning_rate": 3.856365659664399e-05,
      "loss": 0.0225,
      "step": 9716,
      "task_loss": 0.06006855517625809
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0179976187646389,
      "epoch": 9.23,
      "learning_rate": 3.855470468232518e-05,
      "loss": 0.0237,
      "step": 9717,
      "task_loss": 0.07452400773763657
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017134424299001694,
      "epoch": 9.23,
      "learning_rate": 3.854575030572898e-05,
      "loss": 0.0229,
      "step": 9718,
      "task_loss": 0.0750466138124466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015787985175848007,
      "epoch": 9.23,
      "learning_rate": 3.853679346848201e-05,
      "loss": 0.0209,
      "step": 9719,
      "task_loss": 0.06708446890115738
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023376762866973877,
      "epoch": 9.23,
      "learning_rate": 3.8527834172211306e-05,
      "loss": 0.0362,
      "step": 9720,
      "task_loss": 0.15146197378635406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017823493108153343,
      "epoch": 9.23,
      "learning_rate": 3.851887241854438e-05,
      "loss": 0.0172,
      "step": 9721,
      "task_loss": 0.011508205905556679
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01219139527529478,
      "epoch": 9.23,
      "learning_rate": 3.850990820910917e-05,
      "loss": 0.0114,
      "step": 9722,
      "task_loss": 0.003914957866072655
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021136948838829994,
      "epoch": 9.23,
      "learning_rate": 3.8500941545534065e-05,
      "loss": 0.0259,
      "step": 9723,
      "task_loss": 0.06834466755390167
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05498252436518669,
      "epoch": 9.23,
      "learning_rate": 3.849197242944791e-05,
      "loss": 0.0571,
      "step": 9724,
      "task_loss": 0.07627555727958679
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03645971044898033,
      "epoch": 9.24,
      "learning_rate": 3.8483000862479986e-05,
      "loss": 0.0411,
      "step": 9725,
      "task_loss": 0.08248203247785568
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023510009050369263,
      "epoch": 9.24,
      "learning_rate": 3.8474026846260015e-05,
      "loss": 0.0419,
      "step": 9726,
      "task_loss": 0.20696833729743958
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.037108682096004486,
      "epoch": 9.24,
      "learning_rate": 3.846505038241818e-05,
      "loss": 0.0336,
      "step": 9727,
      "task_loss": 0.0021884366869926453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05788455903530121,
      "epoch": 9.24,
      "learning_rate": 3.84560714725851e-05,
      "loss": 0.0531,
      "step": 9728,
      "task_loss": 0.009828178212046623
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01626589521765709,
      "epoch": 9.24,
      "learning_rate": 3.8447090118391814e-05,
      "loss": 0.0339,
      "step": 9729,
      "task_loss": 0.19226785004138947
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.044995106756687164,
      "epoch": 9.24,
      "learning_rate": 3.8438106321469864e-05,
      "loss": 0.0445,
      "step": 9730,
      "task_loss": 0.03956909477710724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018232377246022224,
      "epoch": 9.24,
      "learning_rate": 3.842912008345117e-05,
      "loss": 0.0229,
      "step": 9731,
      "task_loss": 0.0647912323474884
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01168222725391388,
      "epoch": 9.24,
      "learning_rate": 3.842013140596815e-05,
      "loss": 0.0143,
      "step": 9732,
      "task_loss": 0.0380130261182785
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011876760981976986,
      "epoch": 9.24,
      "learning_rate": 3.841114029065362e-05,
      "loss": 0.011,
      "step": 9733,
      "task_loss": 0.003131164237856865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05818028748035431,
      "epoch": 9.24,
      "learning_rate": 3.8402146739140874e-05,
      "loss": 0.0713,
      "step": 9734,
      "task_loss": 0.1889064908027649
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016233699396252632,
      "epoch": 9.25,
      "learning_rate": 3.8393150753063614e-05,
      "loss": 0.0202,
      "step": 9735,
      "task_loss": 0.05590314790606499
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06847208738327026,
      "epoch": 9.25,
      "learning_rate": 3.838415233405603e-05,
      "loss": 0.0667,
      "step": 9736,
      "task_loss": 0.050349440425634384
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022901874035596848,
      "epoch": 9.25,
      "learning_rate": 3.837515148375271e-05,
      "loss": 0.021,
      "step": 9737,
      "task_loss": 0.003996755927801132
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02838345803320408,
      "epoch": 9.25,
      "learning_rate": 3.836614820378871e-05,
      "loss": 0.0341,
      "step": 9738,
      "task_loss": 0.08570387959480286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016883879899978638,
      "epoch": 9.25,
      "learning_rate": 3.835714249579952e-05,
      "loss": 0.0167,
      "step": 9739,
      "task_loss": 0.015314383432269096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03494098410010338,
      "epoch": 9.25,
      "learning_rate": 3.8348134361421064e-05,
      "loss": 0.0337,
      "step": 9740,
      "task_loss": 0.022901371121406555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016907619312405586,
      "epoch": 9.25,
      "learning_rate": 3.8339123802289716e-05,
      "loss": 0.0157,
      "step": 9741,
      "task_loss": 0.00507272407412529
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.043327562510967255,
      "epoch": 9.25,
      "learning_rate": 3.8330110820042285e-05,
      "loss": 0.0487,
      "step": 9742,
      "task_loss": 0.09731481969356537
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12834414839744568,
      "epoch": 9.25,
      "learning_rate": 3.8321095416316024e-05,
      "loss": 0.136,
      "step": 9743,
      "task_loss": 0.20463687181472778
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.032464127987623215,
      "epoch": 9.25,
      "learning_rate": 3.831207759274863e-05,
      "loss": 0.0303,
      "step": 9744,
      "task_loss": 0.011053359135985374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022374771535396576,
      "epoch": 9.25,
      "learning_rate": 3.8303057350978224e-05,
      "loss": 0.0216,
      "step": 9745,
      "task_loss": 0.014581706374883652
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.045660026371479034,
      "epoch": 9.26,
      "learning_rate": 3.829403469264339e-05,
      "loss": 0.0425,
      "step": 9746,
      "task_loss": 0.013828214257955551
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023104814812541008,
      "epoch": 9.26,
      "learning_rate": 3.828500961938313e-05,
      "loss": 0.0377,
      "step": 9747,
      "task_loss": 0.1689058542251587
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010014375671744347,
      "epoch": 9.26,
      "learning_rate": 3.827598213283688e-05,
      "loss": 0.0095,
      "step": 9748,
      "task_loss": 0.004870768636465073
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11848193407058716,
      "epoch": 9.26,
      "learning_rate": 3.8266952234644545e-05,
      "loss": 0.1292,
      "step": 9749,
      "task_loss": 0.22540438175201416
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12723466753959656,
      "epoch": 9.26,
      "learning_rate": 3.825791992644644e-05,
      "loss": 0.1362,
      "step": 9750,
      "task_loss": 0.21650384366512299
    },
    {
      "epoch": 9.26,
      "eval_accuracy": 0.9013761467889908,
      "eval_loss": 0.4256434738636017,
      "eval_runtime": 17.9197,
      "eval_samples_per_second": 48.662,
      "eval_steps_per_second": 6.083,
      "step": 9750
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02083965763449669,
      "epoch": 9.26,
      "learning_rate": 3.824888520988333e-05,
      "loss": 0.0192,
      "step": 9751,
      "task_loss": 0.004488172009587288
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025724343955516815,
      "epoch": 9.26,
      "learning_rate": 3.823984808659641e-05,
      "loss": 0.0374,
      "step": 9752,
      "task_loss": 0.14217713475227356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012440409511327744,
      "epoch": 9.26,
      "learning_rate": 3.8230808558227335e-05,
      "loss": 0.0116,
      "step": 9753,
      "task_loss": 0.0036589261144399643
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009372575208544731,
      "epoch": 9.26,
      "learning_rate": 3.8221766626418155e-05,
      "loss": 0.009,
      "step": 9754,
      "task_loss": 0.005342619493603706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.032062187790870667,
      "epoch": 9.26,
      "learning_rate": 3.821272229281139e-05,
      "loss": 0.0349,
      "step": 9755,
      "task_loss": 0.06090862303972244
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05444857478141785,
      "epoch": 9.26,
      "learning_rate": 3.820367555904999e-05,
      "loss": 0.0574,
      "step": 9756,
      "task_loss": 0.08438717573881149
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02243727631866932,
      "epoch": 9.27,
      "learning_rate": 3.819462642677733e-05,
      "loss": 0.0375,
      "step": 9757,
      "task_loss": 0.17298638820648193
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019842922687530518,
      "epoch": 9.27,
      "learning_rate": 3.818557489763724e-05,
      "loss": 0.0347,
      "step": 9758,
      "task_loss": 0.16849234700202942
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014697995036840439,
      "epoch": 9.27,
      "learning_rate": 3.817652097327396e-05,
      "loss": 0.0137,
      "step": 9759,
      "task_loss": 0.004293628036975861
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0161375030875206,
      "epoch": 9.27,
      "learning_rate": 3.81674646553322e-05,
      "loss": 0.0251,
      "step": 9760,
      "task_loss": 0.10576816648244858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025952016934752464,
      "epoch": 9.27,
      "learning_rate": 3.815840594545706e-05,
      "loss": 0.0432,
      "step": 9761,
      "task_loss": 0.19875141978263855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030038634315133095,
      "epoch": 9.27,
      "learning_rate": 3.814934484529411e-05,
      "loss": 0.0387,
      "step": 9762,
      "task_loss": 0.11658079922199249
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1559152901172638,
      "epoch": 9.27,
      "learning_rate": 3.8140281356489346e-05,
      "loss": 0.1487,
      "step": 9763,
      "task_loss": 0.08386341482400894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02892562374472618,
      "epoch": 9.27,
      "learning_rate": 3.8131215480689184e-05,
      "loss": 0.0478,
      "step": 9764,
      "task_loss": 0.21779313683509827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021166235208511353,
      "epoch": 9.27,
      "learning_rate": 3.812214721954049e-05,
      "loss": 0.0218,
      "step": 9765,
      "task_loss": 0.027633680030703545
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021776802837848663,
      "epoch": 9.27,
      "learning_rate": 3.811307657469055e-05,
      "loss": 0.0203,
      "step": 9766,
      "task_loss": 0.007460303604602814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022118225693702698,
      "epoch": 9.28,
      "learning_rate": 3.8104003547787105e-05,
      "loss": 0.0211,
      "step": 9767,
      "task_loss": 0.011582162231206894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012279561720788479,
      "epoch": 9.28,
      "learning_rate": 3.809492814047831e-05,
      "loss": 0.0181,
      "step": 9768,
      "task_loss": 0.07050621509552002
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022169344127178192,
      "epoch": 9.28,
      "learning_rate": 3.8085850354412745e-05,
      "loss": 0.0205,
      "step": 9769,
      "task_loss": 0.005162643268704414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.032863132655620575,
      "epoch": 9.28,
      "learning_rate": 3.807677019123944e-05,
      "loss": 0.0387,
      "step": 9770,
      "task_loss": 0.09106673300266266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0191971268504858,
      "epoch": 9.28,
      "learning_rate": 3.806768765260785e-05,
      "loss": 0.021,
      "step": 9771,
      "task_loss": 0.037707261741161346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01197386346757412,
      "epoch": 9.28,
      "learning_rate": 3.805860274016787e-05,
      "loss": 0.0118,
      "step": 9772,
      "task_loss": 0.00973932072520256
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024311240762472153,
      "epoch": 9.28,
      "learning_rate": 3.8049515455569816e-05,
      "loss": 0.0404,
      "step": 9773,
      "task_loss": 0.1852773129940033
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01635361835360527,
      "epoch": 9.28,
      "learning_rate": 3.804042580046442e-05,
      "loss": 0.0222,
      "step": 9774,
      "task_loss": 0.07465977221727371
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023430045694112778,
      "epoch": 9.28,
      "learning_rate": 3.803133377650288e-05,
      "loss": 0.0294,
      "step": 9775,
      "task_loss": 0.08316051959991455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01706859841942787,
      "epoch": 9.28,
      "learning_rate": 3.80222393853368e-05,
      "loss": 0.0314,
      "step": 9776,
      "task_loss": 0.1608736664056778
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01352263055741787,
      "epoch": 9.28,
      "learning_rate": 3.801314262861822e-05,
      "loss": 0.0127,
      "step": 9777,
      "task_loss": 0.005207255482673645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.039057813584804535,
      "epoch": 9.29,
      "learning_rate": 3.800404350799961e-05,
      "loss": 0.036,
      "step": 9778,
      "task_loss": 0.008711079135537148
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016933217644691467,
      "epoch": 9.29,
      "learning_rate": 3.799494202513386e-05,
      "loss": 0.0157,
      "step": 9779,
      "task_loss": 0.004555156454443932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07299914956092834,
      "epoch": 9.29,
      "learning_rate": 3.798583818167432e-05,
      "loss": 0.0931,
      "step": 9780,
      "task_loss": 0.27392470836639404
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012407921254634857,
      "epoch": 9.29,
      "learning_rate": 3.797673197927473e-05,
      "loss": 0.0119,
      "step": 9781,
      "task_loss": 0.006941312924027443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019861984997987747,
      "epoch": 9.29,
      "learning_rate": 3.796762341958927e-05,
      "loss": 0.0185,
      "step": 9782,
      "task_loss": 0.0062296707183122635
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03457757458090782,
      "epoch": 9.29,
      "learning_rate": 3.795851250427257e-05,
      "loss": 0.0449,
      "step": 9783,
      "task_loss": 0.1378023773431778
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07008830457925797,
      "epoch": 9.29,
      "learning_rate": 3.794939923497967e-05,
      "loss": 0.0876,
      "step": 9784,
      "task_loss": 0.2450743168592453
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020512813702225685,
      "epoch": 9.29,
      "learning_rate": 3.794028361336603e-05,
      "loss": 0.0237,
      "step": 9785,
      "task_loss": 0.05278385803103447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05949783697724342,
      "epoch": 9.29,
      "learning_rate": 3.793116564108754e-05,
      "loss": 0.0719,
      "step": 9786,
      "task_loss": 0.1833191066980362
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012088644318282604,
      "epoch": 9.29,
      "learning_rate": 3.7922045319800545e-05,
      "loss": 0.0161,
      "step": 9787,
      "task_loss": 0.052241381257772446
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0391070693731308,
      "epoch": 9.3,
      "learning_rate": 3.7912922651161783e-05,
      "loss": 0.0505,
      "step": 9788,
      "task_loss": 0.15337170660495758
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01692778989672661,
      "epoch": 9.3,
      "learning_rate": 3.790379763682844e-05,
      "loss": 0.0157,
      "step": 9789,
      "task_loss": 0.004635356366634369
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.15806353092193604,
      "epoch": 9.3,
      "learning_rate": 3.7894670278458096e-05,
      "loss": 0.152,
      "step": 9790,
      "task_loss": 0.09754717350006104
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12203462421894073,
      "epoch": 9.3,
      "learning_rate": 3.7885540577708804e-05,
      "loss": 0.1233,
      "step": 9791,
      "task_loss": 0.13422806560993195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07689625769853592,
      "epoch": 9.3,
      "learning_rate": 3.7876408536239006e-05,
      "loss": 0.0828,
      "step": 9792,
      "task_loss": 0.1364186704158783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03227274864912033,
      "epoch": 9.3,
      "learning_rate": 3.7867274155707585e-05,
      "loss": 0.0374,
      "step": 9793,
      "task_loss": 0.08388587832450867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03347526490688324,
      "epoch": 9.3,
      "learning_rate": 3.7858137437773845e-05,
      "loss": 0.0504,
      "step": 9794,
      "task_loss": 0.2030281275510788
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02590757980942726,
      "epoch": 9.3,
      "learning_rate": 3.784899838409751e-05,
      "loss": 0.0291,
      "step": 9795,
      "task_loss": 0.05813925340771675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024779632687568665,
      "epoch": 9.3,
      "learning_rate": 3.783985699633874e-05,
      "loss": 0.0318,
      "step": 9796,
      "task_loss": 0.09515906125307083
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12447907775640488,
      "epoch": 9.3,
      "learning_rate": 3.783071327615811e-05,
      "loss": 0.1319,
      "step": 9797,
      "task_loss": 0.19891595840454102
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06262899935245514,
      "epoch": 9.3,
      "learning_rate": 3.7821567225216615e-05,
      "loss": 0.0599,
      "step": 9798,
      "task_loss": 0.035038989037275314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03002050705254078,
      "epoch": 9.31,
      "learning_rate": 3.781241884517569e-05,
      "loss": 0.0338,
      "step": 9799,
      "task_loss": 0.06747624278068542
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04927913099527359,
      "epoch": 9.31,
      "learning_rate": 3.780326813769717e-05,
      "loss": 0.0507,
      "step": 9800,
      "task_loss": 0.06330694258213043
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03860379010438919,
      "epoch": 9.31,
      "learning_rate": 3.779411510444334e-05,
      "loss": 0.0354,
      "step": 9801,
      "task_loss": 0.006218938156962395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028020869940519333,
      "epoch": 9.31,
      "learning_rate": 3.778495974707688e-05,
      "loss": 0.0354,
      "step": 9802,
      "task_loss": 0.10227955132722855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008688807487487793,
      "epoch": 9.31,
      "learning_rate": 3.7775802067260905e-05,
      "loss": 0.0106,
      "step": 9803,
      "task_loss": 0.02761073224246502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.15427309274673462,
      "epoch": 9.31,
      "learning_rate": 3.776664206665896e-05,
      "loss": 0.1546,
      "step": 9804,
      "task_loss": 0.15730072557926178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02751828357577324,
      "epoch": 9.31,
      "learning_rate": 3.7757479746935e-05,
      "loss": 0.0351,
      "step": 9805,
      "task_loss": 0.10342276096343994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02257220633327961,
      "epoch": 9.31,
      "learning_rate": 3.77483151097534e-05,
      "loss": 0.0333,
      "step": 9806,
      "task_loss": 0.1300220787525177
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.16032102704048157,
      "epoch": 9.31,
      "learning_rate": 3.773914815677897e-05,
      "loss": 0.154,
      "step": 9807,
      "task_loss": 0.09661975502967834
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.039088841527700424,
      "epoch": 9.31,
      "learning_rate": 3.7729978889676915e-05,
      "loss": 0.0461,
      "step": 9808,
      "task_loss": 0.10870218276977539
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020086947828531265,
      "epoch": 9.32,
      "learning_rate": 3.7720807310112896e-05,
      "loss": 0.0218,
      "step": 9809,
      "task_loss": 0.03744789958000183
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012433727271854877,
      "epoch": 9.32,
      "learning_rate": 3.7711633419752954e-05,
      "loss": 0.0208,
      "step": 9810,
      "task_loss": 0.09626266360282898
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020316528156399727,
      "epoch": 9.32,
      "learning_rate": 3.7702457220263595e-05,
      "loss": 0.0277,
      "step": 9811,
      "task_loss": 0.09421400725841522
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06558854132890701,
      "epoch": 9.32,
      "learning_rate": 3.76932787133117e-05,
      "loss": 0.0822,
      "step": 9812,
      "task_loss": 0.23205044865608215
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05739504098892212,
      "epoch": 9.32,
      "learning_rate": 3.768409790056459e-05,
      "loss": 0.0695,
      "step": 9813,
      "task_loss": 0.1781667172908783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09154346585273743,
      "epoch": 9.32,
      "learning_rate": 3.7674914783690006e-05,
      "loss": 0.0877,
      "step": 9814,
      "task_loss": 0.05328264459967613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04409680515527725,
      "epoch": 9.32,
      "learning_rate": 3.7665729364356115e-05,
      "loss": 0.0559,
      "step": 9815,
      "task_loss": 0.16190242767333984
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027593711391091347,
      "epoch": 9.32,
      "learning_rate": 3.7656541644231494e-05,
      "loss": 0.0265,
      "step": 9816,
      "task_loss": 0.01616600714623928
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.042008671909570694,
      "epoch": 9.32,
      "learning_rate": 3.764735162498512e-05,
      "loss": 0.0398,
      "step": 9817,
      "task_loss": 0.01970026269555092
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08180496841669083,
      "epoch": 9.32,
      "learning_rate": 3.763815930828641e-05,
      "loss": 0.0787,
      "step": 9818,
      "task_loss": 0.05091682821512222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019493555650115013,
      "epoch": 9.32,
      "learning_rate": 3.76289646958052e-05,
      "loss": 0.0245,
      "step": 9819,
      "task_loss": 0.06980308145284653
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04548114538192749,
      "epoch": 9.33,
      "learning_rate": 3.761976778921173e-05,
      "loss": 0.0504,
      "step": 9820,
      "task_loss": 0.09444499015808105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0342935174703598,
      "epoch": 9.33,
      "learning_rate": 3.761056859017667e-05,
      "loss": 0.0368,
      "step": 9821,
      "task_loss": 0.05982755869626999
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029260210692882538,
      "epoch": 9.33,
      "learning_rate": 3.7601367100371085e-05,
      "loss": 0.0385,
      "step": 9822,
      "task_loss": 0.12197436392307281
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02063072845339775,
      "epoch": 9.33,
      "learning_rate": 3.759216332146649e-05,
      "loss": 0.0332,
      "step": 9823,
      "task_loss": 0.1467389613389969
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026403427124023438,
      "epoch": 9.33,
      "learning_rate": 3.7582957255134765e-05,
      "loss": 0.0356,
      "step": 9824,
      "task_loss": 0.11809330433607101
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009067846462130547,
      "epoch": 9.33,
      "learning_rate": 3.7573748903048266e-05,
      "loss": 0.0086,
      "step": 9825,
      "task_loss": 0.004426542669534683
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0586862787604332,
      "epoch": 9.33,
      "learning_rate": 3.756453826687972e-05,
      "loss": 0.0689,
      "step": 9826,
      "task_loss": 0.1604919135570526
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1495116949081421,
      "epoch": 9.33,
      "learning_rate": 3.755532534830229e-05,
      "loss": 0.1435,
      "step": 9827,
      "task_loss": 0.08945684134960175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.056340716779232025,
      "epoch": 9.33,
      "learning_rate": 3.7546110148989535e-05,
      "loss": 0.0687,
      "step": 9828,
      "task_loss": 0.1797863095998764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03908500820398331,
      "epoch": 9.33,
      "learning_rate": 3.7536892670615454e-05,
      "loss": 0.044,
      "step": 9829,
      "task_loss": 0.0877419114112854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012538935989141464,
      "epoch": 9.34,
      "learning_rate": 3.752767291485444e-05,
      "loss": 0.0195,
      "step": 9830,
      "task_loss": 0.08169474452733994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0177429411560297,
      "epoch": 9.34,
      "learning_rate": 3.7518450883381306e-05,
      "loss": 0.017,
      "step": 9831,
      "task_loss": 0.010471796616911888
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012262466363608837,
      "epoch": 9.34,
      "learning_rate": 3.750922657787128e-05,
      "loss": 0.0298,
      "step": 9832,
      "task_loss": 0.18793505430221558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12192486971616745,
      "epoch": 9.34,
      "learning_rate": 3.7500000000000003e-05,
      "loss": 0.1285,
      "step": 9833,
      "task_loss": 0.18805371224880219
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01544223167002201,
      "epoch": 9.34,
      "learning_rate": 3.7490771151443525e-05,
      "loss": 0.0146,
      "step": 9834,
      "task_loss": 0.006540972739458084
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028887242078781128,
      "epoch": 9.34,
      "learning_rate": 3.748154003387831e-05,
      "loss": 0.0273,
      "step": 9835,
      "task_loss": 0.01312171295285225
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014016509056091309,
      "epoch": 9.34,
      "learning_rate": 3.7472306648981235e-05,
      "loss": 0.0132,
      "step": 9836,
      "task_loss": 0.0060823168605566025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.059632837772369385,
      "epoch": 9.34,
      "learning_rate": 3.746307099842959e-05,
      "loss": 0.055,
      "step": 9837,
      "task_loss": 0.01375570334494114
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008659922517836094,
      "epoch": 9.34,
      "learning_rate": 3.745383308390108e-05,
      "loss": 0.0153,
      "step": 9838,
      "task_loss": 0.07544966042041779
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.038622692227363586,
      "epoch": 9.34,
      "learning_rate": 3.74445929070738e-05,
      "loss": 0.0406,
      "step": 9839,
      "task_loss": 0.05818815529346466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05248776823282242,
      "epoch": 9.34,
      "learning_rate": 3.74353504696263e-05,
      "loss": 0.054,
      "step": 9840,
      "task_loss": 0.06735575944185257
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0463816337287426,
      "epoch": 9.35,
      "learning_rate": 3.742610577323749e-05,
      "loss": 0.0579,
      "step": 9841,
      "task_loss": 0.16190406680107117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04839491844177246,
      "epoch": 9.35,
      "learning_rate": 3.7416858819586724e-05,
      "loss": 0.0574,
      "step": 9842,
      "task_loss": 0.1388835310935974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0757160484790802,
      "epoch": 9.35,
      "learning_rate": 3.740760961035375e-05,
      "loss": 0.0747,
      "step": 9843,
      "task_loss": 0.06535409390926361
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02673790417611599,
      "epoch": 9.35,
      "learning_rate": 3.739835814721874e-05,
      "loss": 0.0248,
      "step": 9844,
      "task_loss": 0.007372252643108368
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018942799419164658,
      "epoch": 9.35,
      "learning_rate": 3.738910443186226e-05,
      "loss": 0.0175,
      "step": 9845,
      "task_loss": 0.004348035901784897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020467443391680717,
      "epoch": 9.35,
      "learning_rate": 3.737984846596528e-05,
      "loss": 0.0203,
      "step": 9846,
      "task_loss": 0.018603015691041946
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020978864282369614,
      "epoch": 9.35,
      "learning_rate": 3.737059025120922e-05,
      "loss": 0.02,
      "step": 9847,
      "task_loss": 0.01105569303035736
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06590712815523148,
      "epoch": 9.35,
      "learning_rate": 3.7361329789275855e-05,
      "loss": 0.0767,
      "step": 9848,
      "task_loss": 0.1739407181739807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0286872461438179,
      "epoch": 9.35,
      "learning_rate": 3.7352067081847405e-05,
      "loss": 0.0266,
      "step": 9849,
      "task_loss": 0.007965076714754105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08515746146440506,
      "epoch": 9.35,
      "learning_rate": 3.734280213060649e-05,
      "loss": 0.0818,
      "step": 9850,
      "task_loss": 0.05148651823401451
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02829767018556595,
      "epoch": 9.36,
      "learning_rate": 3.7333534937236105e-05,
      "loss": 0.0399,
      "step": 9851,
      "task_loss": 0.14469148218631744
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0619497075676918,
      "epoch": 9.36,
      "learning_rate": 3.7324265503419716e-05,
      "loss": 0.0694,
      "step": 9852,
      "task_loss": 0.13681992888450623
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02087024413049221,
      "epoch": 9.36,
      "learning_rate": 3.731499383084114e-05,
      "loss": 0.0201,
      "step": 9853,
      "task_loss": 0.012817522510886192
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05007238686084747,
      "epoch": 9.36,
      "learning_rate": 3.730571992118462e-05,
      "loss": 0.0491,
      "step": 9854,
      "task_loss": 0.040530577301979065
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01564808376133442,
      "epoch": 9.36,
      "learning_rate": 3.7296443776134814e-05,
      "loss": 0.0155,
      "step": 9855,
      "task_loss": 0.013979200273752213
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01755111664533615,
      "epoch": 9.36,
      "learning_rate": 3.7287165397376775e-05,
      "loss": 0.0163,
      "step": 9856,
      "task_loss": 0.004699693992733955
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09714722633361816,
      "epoch": 9.36,
      "learning_rate": 3.727788478659597e-05,
      "loss": 0.106,
      "step": 9857,
      "task_loss": 0.18595141172409058
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0596044659614563,
      "epoch": 9.36,
      "learning_rate": 3.726860194547826e-05,
      "loss": 0.0652,
      "step": 9858,
      "task_loss": 0.11533902585506439
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009165867231786251,
      "epoch": 9.36,
      "learning_rate": 3.725931687570992e-05,
      "loss": 0.0224,
      "step": 9859,
      "task_loss": 0.1417846381664276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008022119291126728,
      "epoch": 9.36,
      "learning_rate": 3.7250029578977625e-05,
      "loss": 0.0076,
      "step": 9860,
      "task_loss": 0.004271337762475014
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1454535573720932,
      "epoch": 9.36,
      "learning_rate": 3.724074005696847e-05,
      "loss": 0.1585,
      "step": 9861,
      "task_loss": 0.2754858136177063
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0446578785777092,
      "epoch": 9.37,
      "learning_rate": 3.723144831136992e-05,
      "loss": 0.0508,
      "step": 9862,
      "task_loss": 0.10627266019582748
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009835630655288696,
      "epoch": 9.37,
      "learning_rate": 3.722215434386988e-05,
      "loss": 0.0124,
      "step": 9863,
      "task_loss": 0.03548569977283478
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07582224905490875,
      "epoch": 9.37,
      "learning_rate": 3.721285815615665e-05,
      "loss": 0.088,
      "step": 9864,
      "task_loss": 0.19718605279922485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02364090085029602,
      "epoch": 9.37,
      "learning_rate": 3.7203559749918904e-05,
      "loss": 0.0319,
      "step": 9865,
      "task_loss": 0.10592564195394516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06448841094970703,
      "epoch": 9.37,
      "learning_rate": 3.7194259126845764e-05,
      "loss": 0.0599,
      "step": 9866,
      "task_loss": 0.018569234758615494
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018728742375969887,
      "epoch": 9.37,
      "learning_rate": 3.7184956288626724e-05,
      "loss": 0.0172,
      "step": 9867,
      "task_loss": 0.00367145799100399
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0083607267588377,
      "epoch": 9.37,
      "learning_rate": 3.71756512369517e-05,
      "loss": 0.008,
      "step": 9868,
      "task_loss": 0.004354575648903847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02431337907910347,
      "epoch": 9.37,
      "learning_rate": 3.716634397351097e-05,
      "loss": 0.0261,
      "step": 9869,
      "task_loss": 0.04231826961040497
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028572488576173782,
      "epoch": 9.37,
      "learning_rate": 3.715703449999528e-05,
      "loss": 0.0353,
      "step": 9870,
      "task_loss": 0.09551827609539032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015134245157241821,
      "epoch": 9.37,
      "learning_rate": 3.7147722818095724e-05,
      "loss": 0.0236,
      "step": 9871,
      "task_loss": 0.09958845376968384
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01670273393392563,
      "epoch": 9.38,
      "learning_rate": 3.713840892950381e-05,
      "loss": 0.0327,
      "step": 9872,
      "task_loss": 0.17647552490234375
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07219713926315308,
      "epoch": 9.38,
      "learning_rate": 3.712909283591145e-05,
      "loss": 0.0823,
      "step": 9873,
      "task_loss": 0.17350198328495026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0580780953168869,
      "epoch": 9.38,
      "learning_rate": 3.7119774539010967e-05,
      "loss": 0.0533,
      "step": 9874,
      "task_loss": 0.010585742071270943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03432053327560425,
      "epoch": 9.38,
      "learning_rate": 3.711045404049507e-05,
      "loss": 0.0325,
      "step": 9875,
      "task_loss": 0.016367098316550255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014983313158154488,
      "epoch": 9.38,
      "learning_rate": 3.710113134205687e-05,
      "loss": 0.0138,
      "step": 9876,
      "task_loss": 0.002722267061471939
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04396282881498337,
      "epoch": 9.38,
      "learning_rate": 3.709180644538988e-05,
      "loss": 0.048,
      "step": 9877,
      "task_loss": 0.08390352874994278
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014825180172920227,
      "epoch": 9.38,
      "learning_rate": 3.708247935218802e-05,
      "loss": 0.0202,
      "step": 9878,
      "task_loss": 0.06829116493463516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1081385537981987,
      "epoch": 9.38,
      "learning_rate": 3.707315006414559e-05,
      "loss": 0.1184,
      "step": 9879,
      "task_loss": 0.21026736497879028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06750968843698502,
      "epoch": 9.38,
      "learning_rate": 3.706381858295731e-05,
      "loss": 0.0675,
      "step": 9880,
      "task_loss": 0.06727669388055801
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02159801311790943,
      "epoch": 9.38,
      "learning_rate": 3.705448491031829e-05,
      "loss": 0.0277,
      "step": 9881,
      "task_loss": 0.08293137699365616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07070346176624298,
      "epoch": 9.38,
      "learning_rate": 3.7045149047924016e-05,
      "loss": 0.071,
      "step": 9882,
      "task_loss": 0.07330311834812164
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014572136104106903,
      "epoch": 9.39,
      "learning_rate": 3.703581099747041e-05,
      "loss": 0.0231,
      "step": 9883,
      "task_loss": 0.09984834492206573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.055414482951164246,
      "epoch": 9.39,
      "learning_rate": 3.702647076065378e-05,
      "loss": 0.0554,
      "step": 9884,
      "task_loss": 0.055326469242572784
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025832070037722588,
      "epoch": 9.39,
      "learning_rate": 3.701712833917082e-05,
      "loss": 0.035,
      "step": 9885,
      "task_loss": 0.1178750991821289
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013981124386191368,
      "epoch": 9.39,
      "learning_rate": 3.700778373471861e-05,
      "loss": 0.0131,
      "step": 9886,
      "task_loss": 0.005626115947961807
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024125682190060616,
      "epoch": 9.39,
      "learning_rate": 3.699843694899467e-05,
      "loss": 0.0283,
      "step": 9887,
      "task_loss": 0.06636115163564682
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022186098620295525,
      "epoch": 9.39,
      "learning_rate": 3.698908798369686e-05,
      "loss": 0.0206,
      "step": 9888,
      "task_loss": 0.006147833541035652
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09722729027271271,
      "epoch": 9.39,
      "learning_rate": 3.697973684052347e-05,
      "loss": 0.1004,
      "step": 9889,
      "task_loss": 0.12891815602779388
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.006281149573624134,
      "epoch": 9.39,
      "learning_rate": 3.697038352117321e-05,
      "loss": 0.0182,
      "step": 9890,
      "task_loss": 0.12592127919197083
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014334081672132015,
      "epoch": 9.39,
      "learning_rate": 3.6961028027345114e-05,
      "loss": 0.024,
      "step": 9891,
      "task_loss": 0.11108017712831497
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07208122313022614,
      "epoch": 9.39,
      "learning_rate": 3.695167036073868e-05,
      "loss": 0.0769,
      "step": 9892,
      "task_loss": 0.12063595652580261
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04806242883205414,
      "epoch": 9.4,
      "learning_rate": 3.694231052305376e-05,
      "loss": 0.0582,
      "step": 9893,
      "task_loss": 0.14957112073898315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022140424698591232,
      "epoch": 9.4,
      "learning_rate": 3.693294851599063e-05,
      "loss": 0.0354,
      "step": 9894,
      "task_loss": 0.1551881581544876
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017354466021060944,
      "epoch": 9.4,
      "learning_rate": 3.692358434124992e-05,
      "loss": 0.021,
      "step": 9895,
      "task_loss": 0.05346723645925522
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02076183259487152,
      "epoch": 9.4,
      "learning_rate": 3.69142180005327e-05,
      "loss": 0.0196,
      "step": 9896,
      "task_loss": 0.009420402348041534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0411936454474926,
      "epoch": 9.4,
      "learning_rate": 3.69048494955404e-05,
      "loss": 0.0466,
      "step": 9897,
      "task_loss": 0.09573011100292206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03919178247451782,
      "epoch": 9.4,
      "learning_rate": 3.689547882797485e-05,
      "loss": 0.0358,
      "step": 9898,
      "task_loss": 0.005710486322641373
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0696905106306076,
      "epoch": 9.4,
      "learning_rate": 3.688610599953828e-05,
      "loss": 0.0671,
      "step": 9899,
      "task_loss": 0.0438590869307518
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07281279563903809,
      "epoch": 9.4,
      "learning_rate": 3.6876731011933316e-05,
      "loss": 0.0817,
      "step": 9900,
      "task_loss": 0.1620171219110489
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016692060977220535,
      "epoch": 9.4,
      "learning_rate": 3.686735386686296e-05,
      "loss": 0.0153,
      "step": 9901,
      "task_loss": 0.002423325553536415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03054118901491165,
      "epoch": 9.4,
      "learning_rate": 3.685797456603062e-05,
      "loss": 0.0343,
      "step": 9902,
      "task_loss": 0.06791390478610992
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013188479468226433,
      "epoch": 9.4,
      "learning_rate": 3.684859311114009e-05,
      "loss": 0.0183,
      "step": 9903,
      "task_loss": 0.06404251605272293
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030764251947402954,
      "epoch": 9.41,
      "learning_rate": 3.6839209503895566e-05,
      "loss": 0.0468,
      "step": 9904,
      "task_loss": 0.19135428965091705
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024435993283987045,
      "epoch": 9.41,
      "learning_rate": 3.6829823746001616e-05,
      "loss": 0.03,
      "step": 9905,
      "task_loss": 0.08053772151470184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03574752062559128,
      "epoch": 9.41,
      "learning_rate": 3.6820435839163205e-05,
      "loss": 0.04,
      "step": 9906,
      "task_loss": 0.07870674878358841
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13980330526828766,
      "epoch": 9.41,
      "learning_rate": 3.68110457850857e-05,
      "loss": 0.1499,
      "step": 9907,
      "task_loss": 0.24049611389636993
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0633138120174408,
      "epoch": 9.41,
      "learning_rate": 3.680165358547484e-05,
      "loss": 0.0682,
      "step": 9908,
      "task_loss": 0.1119156926870346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03244736045598984,
      "epoch": 9.41,
      "learning_rate": 3.6792259242036776e-05,
      "loss": 0.0359,
      "step": 9909,
      "task_loss": 0.06734733283519745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020193520933389664,
      "epoch": 9.41,
      "learning_rate": 3.678286275647802e-05,
      "loss": 0.0188,
      "step": 9910,
      "task_loss": 0.006022298708558083
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01735522225499153,
      "epoch": 9.41,
      "learning_rate": 3.677346413050551e-05,
      "loss": 0.0171,
      "step": 9911,
      "task_loss": 0.014364780858159065
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03136299550533295,
      "epoch": 9.41,
      "learning_rate": 3.6764063365826525e-05,
      "loss": 0.0348,
      "step": 9912,
      "task_loss": 0.06544046103954315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03162465989589691,
      "epoch": 9.41,
      "learning_rate": 3.675466046414878e-05,
      "loss": 0.0339,
      "step": 9913,
      "task_loss": 0.05433046445250511
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028096985071897507,
      "epoch": 9.42,
      "learning_rate": 3.674525542718035e-05,
      "loss": 0.0265,
      "step": 9914,
      "task_loss": 0.012334998697042465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10660208761692047,
      "epoch": 9.42,
      "learning_rate": 3.6735848256629705e-05,
      "loss": 0.1089,
      "step": 9915,
      "task_loss": 0.12940451502799988
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05087193101644516,
      "epoch": 9.42,
      "learning_rate": 3.6726438954205714e-05,
      "loss": 0.06,
      "step": 9916,
      "task_loss": 0.14205744862556458
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.035260528326034546,
      "epoch": 9.42,
      "learning_rate": 3.6717027521617595e-05,
      "loss": 0.0326,
      "step": 9917,
      "task_loss": 0.009002592414617538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.036381468176841736,
      "epoch": 9.42,
      "learning_rate": 3.6707613960575006e-05,
      "loss": 0.0375,
      "step": 9918,
      "task_loss": 0.048002809286117554
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01758906990289688,
      "epoch": 9.42,
      "learning_rate": 3.669819827278795e-05,
      "loss": 0.0164,
      "step": 9919,
      "task_loss": 0.005666827782988548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05025162547826767,
      "epoch": 9.42,
      "learning_rate": 3.668878045996685e-05,
      "loss": 0.0592,
      "step": 9920,
      "task_loss": 0.13993430137634277
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014506649225950241,
      "epoch": 9.42,
      "learning_rate": 3.667936052382248e-05,
      "loss": 0.0134,
      "step": 9921,
      "task_loss": 0.0035849660634994507
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0744752436876297,
      "epoch": 9.42,
      "learning_rate": 3.666993846606602e-05,
      "loss": 0.0755,
      "step": 9922,
      "task_loss": 0.0845727026462555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.037520673125982285,
      "epoch": 9.42,
      "learning_rate": 3.666051428840904e-05,
      "loss": 0.0409,
      "step": 9923,
      "task_loss": 0.07118186354637146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.007355409674346447,
      "epoch": 9.42,
      "learning_rate": 3.665108799256348e-05,
      "loss": 0.007,
      "step": 9924,
      "task_loss": 0.003476981073617935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012398531660437584,
      "epoch": 9.43,
      "learning_rate": 3.6641659580241665e-05,
      "loss": 0.0276,
      "step": 9925,
      "task_loss": 0.16440680623054504
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10659578442573547,
      "epoch": 9.43,
      "learning_rate": 3.663222905315633e-05,
      "loss": 0.1168,
      "step": 9926,
      "task_loss": 0.20838633179664612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06712566316127777,
      "epoch": 9.43,
      "learning_rate": 3.662279641302056e-05,
      "loss": 0.076,
      "step": 9927,
      "task_loss": 0.15624761581420898
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05237956717610359,
      "epoch": 9.43,
      "learning_rate": 3.6613361661547854e-05,
      "loss": 0.052,
      "step": 9928,
      "task_loss": 0.04885098338127136
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05890313908457756,
      "epoch": 9.43,
      "learning_rate": 3.660392480045206e-05,
      "loss": 0.0631,
      "step": 9929,
      "task_loss": 0.10126684606075287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020466038957238197,
      "epoch": 9.43,
      "learning_rate": 3.659448583144745e-05,
      "loss": 0.0355,
      "step": 9930,
      "task_loss": 0.1712527871131897
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01883137971162796,
      "epoch": 9.43,
      "learning_rate": 3.658504475624865e-05,
      "loss": 0.0175,
      "step": 9931,
      "task_loss": 0.005616925656795502
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019618559628725052,
      "epoch": 9.43,
      "learning_rate": 3.657560157657067e-05,
      "loss": 0.0181,
      "step": 9932,
      "task_loss": 0.003942342475056648
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022781796753406525,
      "epoch": 9.43,
      "learning_rate": 3.656615629412892e-05,
      "loss": 0.0382,
      "step": 9933,
      "task_loss": 0.17709828913211823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021544065326452255,
      "epoch": 9.43,
      "learning_rate": 3.655670891063917e-05,
      "loss": 0.0356,
      "step": 9934,
      "task_loss": 0.1624692678451538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.037996262311935425,
      "epoch": 9.43,
      "learning_rate": 3.6547259427817595e-05,
      "loss": 0.0421,
      "step": 9935,
      "task_loss": 0.07946674525737762
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06063329428434372,
      "epoch": 9.44,
      "learning_rate": 3.6537807847380726e-05,
      "loss": 0.0579,
      "step": 9936,
      "task_loss": 0.03335261717438698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03863832727074623,
      "epoch": 9.44,
      "learning_rate": 3.65283541710455e-05,
      "loss": 0.046,
      "step": 9937,
      "task_loss": 0.11236982047557831
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01804421842098236,
      "epoch": 9.44,
      "learning_rate": 3.6518898400529214e-05,
      "loss": 0.0172,
      "step": 9938,
      "task_loss": 0.009426100179553032
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03564087301492691,
      "epoch": 9.44,
      "learning_rate": 3.650944053754956e-05,
      "loss": 0.0408,
      "step": 9939,
      "task_loss": 0.08679479360580444
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.037948817014694214,
      "epoch": 9.44,
      "learning_rate": 3.6499980583824606e-05,
      "loss": 0.0351,
      "step": 9940,
      "task_loss": 0.009799247607588768
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020474538207054138,
      "epoch": 9.44,
      "learning_rate": 3.64905185410728e-05,
      "loss": 0.0258,
      "step": 9941,
      "task_loss": 0.07416295260190964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07244652509689331,
      "epoch": 9.44,
      "learning_rate": 3.6481054411012946e-05,
      "loss": 0.0706,
      "step": 9942,
      "task_loss": 0.05382921174168587
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03393717110157013,
      "epoch": 9.44,
      "learning_rate": 3.647158819536427e-05,
      "loss": 0.0355,
      "step": 9943,
      "task_loss": 0.04978133365511894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02722102776169777,
      "epoch": 9.44,
      "learning_rate": 3.646211989584635e-05,
      "loss": 0.0334,
      "step": 9944,
      "task_loss": 0.08923061192035675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0162800382822752,
      "epoch": 9.44,
      "learning_rate": 3.645264951417915e-05,
      "loss": 0.0222,
      "step": 9945,
      "task_loss": 0.0759543627500534
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012943320907652378,
      "epoch": 9.45,
      "learning_rate": 3.644317705208301e-05,
      "loss": 0.0122,
      "step": 9946,
      "task_loss": 0.005832251161336899
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06934761255979538,
      "epoch": 9.45,
      "learning_rate": 3.643370251127865e-05,
      "loss": 0.0795,
      "step": 9947,
      "task_loss": 0.17082899808883667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03814961761236191,
      "epoch": 9.45,
      "learning_rate": 3.6424225893487166e-05,
      "loss": 0.0484,
      "step": 9948,
      "task_loss": 0.140329971909523
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03574157506227493,
      "epoch": 9.45,
      "learning_rate": 3.641474720043002e-05,
      "loss": 0.0353,
      "step": 9949,
      "task_loss": 0.03175017610192299
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06168286129832268,
      "epoch": 9.45,
      "learning_rate": 3.6405266433829075e-05,
      "loss": 0.0607,
      "step": 9950,
      "task_loss": 0.05187152698636055
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.033880721777677536,
      "epoch": 9.45,
      "learning_rate": 3.639578359540655e-05,
      "loss": 0.0376,
      "step": 9951,
      "task_loss": 0.07117892801761627
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019045129418373108,
      "epoch": 9.45,
      "learning_rate": 3.638629868688506e-05,
      "loss": 0.0257,
      "step": 9952,
      "task_loss": 0.08584122359752655
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024328162893652916,
      "epoch": 9.45,
      "learning_rate": 3.6376811709987574e-05,
      "loss": 0.0236,
      "step": 9953,
      "task_loss": 0.016677698120474815
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05967606231570244,
      "epoch": 9.45,
      "learning_rate": 3.636732266643745e-05,
      "loss": 0.0675,
      "step": 9954,
      "task_loss": 0.13744193315505981
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09701327234506607,
      "epoch": 9.45,
      "learning_rate": 3.635783155795841e-05,
      "loss": 0.1019,
      "step": 9955,
      "task_loss": 0.14588984847068787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022994887083768845,
      "epoch": 9.45,
      "learning_rate": 3.634833838627458e-05,
      "loss": 0.0332,
      "step": 9956,
      "task_loss": 0.12463461607694626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021844923496246338,
      "epoch": 9.46,
      "learning_rate": 3.6338843153110424e-05,
      "loss": 0.0288,
      "step": 9957,
      "task_loss": 0.09174901992082596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04523449391126633,
      "epoch": 9.46,
      "learning_rate": 3.63293458601908e-05,
      "loss": 0.0436,
      "step": 9958,
      "task_loss": 0.028626611456274986
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015888435766100883,
      "epoch": 9.46,
      "learning_rate": 3.631984650924094e-05,
      "loss": 0.0149,
      "step": 9959,
      "task_loss": 0.006228139623999596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04058516398072243,
      "epoch": 9.46,
      "learning_rate": 3.631034510198643e-05,
      "loss": 0.0452,
      "step": 9960,
      "task_loss": 0.08676454424858093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028549985960125923,
      "epoch": 9.46,
      "learning_rate": 3.630084164015328e-05,
      "loss": 0.0261,
      "step": 9961,
      "task_loss": 0.004079824313521385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05375894159078598,
      "epoch": 9.46,
      "learning_rate": 3.6291336125467814e-05,
      "loss": 0.0524,
      "step": 9962,
      "task_loss": 0.040088847279548645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019988451153039932,
      "epoch": 9.46,
      "learning_rate": 3.628182855965676e-05,
      "loss": 0.025,
      "step": 9963,
      "task_loss": 0.07049809396266937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01588454097509384,
      "epoch": 9.46,
      "learning_rate": 3.627231894444721e-05,
      "loss": 0.0162,
      "step": 9964,
      "task_loss": 0.019161686301231384
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07747796177864075,
      "epoch": 9.46,
      "learning_rate": 3.6262807281566634e-05,
      "loss": 0.0743,
      "step": 9965,
      "task_loss": 0.04522999748587608
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03719233348965645,
      "epoch": 9.46,
      "learning_rate": 3.6253293572742884e-05,
      "loss": 0.0341,
      "step": 9966,
      "task_loss": 0.006006931886076927
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022969527170062065,
      "epoch": 9.47,
      "learning_rate": 3.624377781970416e-05,
      "loss": 0.0218,
      "step": 9967,
      "task_loss": 0.011378584429621696
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02225591614842415,
      "epoch": 9.47,
      "learning_rate": 3.6234260024179033e-05,
      "loss": 0.0206,
      "step": 9968,
      "task_loss": 0.005647040903568268
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015963654965162277,
      "epoch": 9.47,
      "learning_rate": 3.622474018789648e-05,
      "loss": 0.0153,
      "step": 9969,
      "task_loss": 0.009000055491924286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012255464680492878,
      "epoch": 9.47,
      "learning_rate": 3.62152183125858e-05,
      "loss": 0.0114,
      "step": 9970,
      "task_loss": 0.0038295499980449677
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02565714716911316,
      "epoch": 9.47,
      "learning_rate": 3.620569439997671e-05,
      "loss": 0.0497,
      "step": 9971,
      "task_loss": 0.2665744721889496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.033734291791915894,
      "epoch": 9.47,
      "learning_rate": 3.6196168451799266e-05,
      "loss": 0.0376,
      "step": 9972,
      "task_loss": 0.07192590832710266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019625328481197357,
      "epoch": 9.47,
      "learning_rate": 3.618664046978389e-05,
      "loss": 0.0229,
      "step": 9973,
      "task_loss": 0.052226584404706955
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07318352907896042,
      "epoch": 9.47,
      "learning_rate": 3.617711045566141e-05,
      "loss": 0.0663,
      "step": 9974,
      "task_loss": 0.004145057871937752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10104160010814667,
      "epoch": 9.47,
      "learning_rate": 3.616757841116298e-05,
      "loss": 0.0914,
      "step": 9975,
      "task_loss": 0.0048540495336055756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0499558225274086,
      "epoch": 9.47,
      "learning_rate": 3.6158044338020155e-05,
      "loss": 0.0594,
      "step": 9976,
      "task_loss": 0.1440429389476776
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018780533224344254,
      "epoch": 9.47,
      "learning_rate": 3.614850823796483e-05,
      "loss": 0.0174,
      "step": 9977,
      "task_loss": 0.0051146019250154495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021981973201036453,
      "epoch": 9.48,
      "learning_rate": 3.6138970112729296e-05,
      "loss": 0.0208,
      "step": 9978,
      "task_loss": 0.010115953162312508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013006994500756264,
      "epoch": 9.48,
      "learning_rate": 3.612942996404619e-05,
      "loss": 0.0122,
      "step": 9979,
      "task_loss": 0.0053765904158353806
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04389118403196335,
      "epoch": 9.48,
      "learning_rate": 3.611988779364853e-05,
      "loss": 0.0429,
      "step": 9980,
      "task_loss": 0.034226249903440475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.062232695519924164,
      "epoch": 9.48,
      "learning_rate": 3.611034360326971e-05,
      "loss": 0.0623,
      "step": 9981,
      "task_loss": 0.062431663274765015
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08030220866203308,
      "epoch": 9.48,
      "learning_rate": 3.6100797394643455e-05,
      "loss": 0.0799,
      "step": 9982,
      "task_loss": 0.07579746097326279
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09911078959703445,
      "epoch": 9.48,
      "learning_rate": 3.60912491695039e-05,
      "loss": 0.0947,
      "step": 9983,
      "task_loss": 0.054534316062927246
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018304534256458282,
      "epoch": 9.48,
      "learning_rate": 3.608169892958551e-05,
      "loss": 0.0293,
      "step": 9984,
      "task_loss": 0.1278255134820938
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0267410259693861,
      "epoch": 9.48,
      "learning_rate": 3.607214667662314e-05,
      "loss": 0.0337,
      "step": 9985,
      "task_loss": 0.09607288241386414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012940805405378342,
      "epoch": 9.48,
      "learning_rate": 3.606259241235201e-05,
      "loss": 0.0164,
      "step": 9986,
      "task_loss": 0.04748379439115524
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05529266595840454,
      "epoch": 9.48,
      "learning_rate": 3.605303613850768e-05,
      "loss": 0.0547,
      "step": 9987,
      "task_loss": 0.048941053450107574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0491475947201252,
      "epoch": 9.49,
      "learning_rate": 3.604347785682611e-05,
      "loss": 0.0475,
      "step": 9988,
      "task_loss": 0.03230959177017212
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.042645324021577835,
      "epoch": 9.49,
      "learning_rate": 3.60339175690436e-05,
      "loss": 0.0437,
      "step": 9989,
      "task_loss": 0.0532299280166626
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06176585331559181,
      "epoch": 9.49,
      "learning_rate": 3.602435527689683e-05,
      "loss": 0.0598,
      "step": 9990,
      "task_loss": 0.04228825867176056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015652582049369812,
      "epoch": 9.49,
      "learning_rate": 3.6014790982122816e-05,
      "loss": 0.024,
      "step": 9991,
      "task_loss": 0.09889158606529236
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04341176524758339,
      "epoch": 9.49,
      "learning_rate": 3.6005224686458985e-05,
      "loss": 0.0519,
      "step": 9992,
      "task_loss": 0.1284147948026657
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04502232372760773,
      "epoch": 9.49,
      "learning_rate": 3.599565639164308e-05,
      "loss": 0.0422,
      "step": 9993,
      "task_loss": 0.01630707085132599
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029705507680773735,
      "epoch": 9.49,
      "learning_rate": 3.5986086099413234e-05,
      "loss": 0.0288,
      "step": 9994,
      "task_loss": 0.020898183807730675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03598994016647339,
      "epoch": 9.49,
      "learning_rate": 3.597651381150795e-05,
      "loss": 0.034,
      "step": 9995,
      "task_loss": 0.015627581626176834
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03647323325276375,
      "epoch": 9.49,
      "learning_rate": 3.5966939529666056e-05,
      "loss": 0.0337,
      "step": 9996,
      "task_loss": 0.008715398609638214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022679002955555916,
      "epoch": 9.49,
      "learning_rate": 3.595736325562679e-05,
      "loss": 0.0215,
      "step": 9997,
      "task_loss": 0.011273723095655441
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04891244322061539,
      "epoch": 9.49,
      "learning_rate": 3.5947784991129716e-05,
      "loss": 0.0549,
      "step": 9998,
      "task_loss": 0.10830025374889374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023395076394081116,
      "epoch": 9.5,
      "learning_rate": 3.593820473791476e-05,
      "loss": 0.0217,
      "step": 9999,
      "task_loss": 0.006120296195149422
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012700533494353294,
      "epoch": 9.5,
      "learning_rate": 3.5928622497722245e-05,
      "loss": 0.0125,
      "step": 10000,
      "task_loss": 0.010491423308849335
    },
    {
      "epoch": 9.5,
      "eval_accuracy": 0.8944954128440367,
      "eval_loss": 0.4492689073085785,
      "eval_runtime": 18.1708,
      "eval_samples_per_second": 47.989,
      "eval_steps_per_second": 5.999,
      "step": 10000
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07806367427110672,
      "epoch": 9.5,
      "learning_rate": 3.591903827229282e-05,
      "loss": 0.0813,
      "step": 10001,
      "task_loss": 0.11088013648986816
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012855404987931252,
      "epoch": 9.5,
      "learning_rate": 3.590945206336751e-05,
      "loss": 0.0119,
      "step": 10002,
      "task_loss": 0.002988189458847046
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06485391408205032,
      "epoch": 9.5,
      "learning_rate": 3.5899863872687675e-05,
      "loss": 0.0683,
      "step": 10003,
      "task_loss": 0.09926420450210571
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02057585120201111,
      "epoch": 9.5,
      "learning_rate": 3.58902737019951e-05,
      "loss": 0.0203,
      "step": 10004,
      "task_loss": 0.018218940123915672
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03241540864109993,
      "epoch": 9.5,
      "learning_rate": 3.5880681553031835e-05,
      "loss": 0.0318,
      "step": 10005,
      "task_loss": 0.026616670191287994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01615779660642147,
      "epoch": 9.5,
      "learning_rate": 3.5871087427540375e-05,
      "loss": 0.0153,
      "step": 10006,
      "task_loss": 0.007706159725785255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03969818726181984,
      "epoch": 9.5,
      "learning_rate": 3.586149132726353e-05,
      "loss": 0.0375,
      "step": 10007,
      "task_loss": 0.01816389709711075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017227375879883766,
      "epoch": 9.5,
      "learning_rate": 3.585189325394447e-05,
      "loss": 0.0161,
      "step": 10008,
      "task_loss": 0.006397966295480728
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01556732039898634,
      "epoch": 9.51,
      "learning_rate": 3.5842293209326746e-05,
      "loss": 0.0145,
      "step": 10009,
      "task_loss": 0.00453593023121357
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012093789875507355,
      "epoch": 9.51,
      "learning_rate": 3.583269119515423e-05,
      "loss": 0.0182,
      "step": 10010,
      "task_loss": 0.0733606368303299
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029773391783237457,
      "epoch": 9.51,
      "learning_rate": 3.58230872131712e-05,
      "loss": 0.0276,
      "step": 10011,
      "task_loss": 0.00802118144929409
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015580402687191963,
      "epoch": 9.51,
      "learning_rate": 3.581348126512225e-05,
      "loss": 0.0147,
      "step": 10012,
      "task_loss": 0.007100388407707214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01875162497162819,
      "epoch": 9.51,
      "learning_rate": 3.5803873352752343e-05,
      "loss": 0.0177,
      "step": 10013,
      "task_loss": 0.00872032716870308
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07013194262981415,
      "epoch": 9.51,
      "learning_rate": 3.5794263477806816e-05,
      "loss": 0.074,
      "step": 10014,
      "task_loss": 0.10908061265945435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023847414180636406,
      "epoch": 9.51,
      "learning_rate": 3.578465164203134e-05,
      "loss": 0.0376,
      "step": 10015,
      "task_loss": 0.16116636991500854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010446256026625633,
      "epoch": 9.51,
      "learning_rate": 3.577503784717195e-05,
      "loss": 0.0099,
      "step": 10016,
      "task_loss": 0.005132999271154404
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.046971529722213745,
      "epoch": 9.51,
      "learning_rate": 3.576542209497505e-05,
      "loss": 0.0496,
      "step": 10017,
      "task_loss": 0.0734739899635315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020745940506458282,
      "epoch": 9.51,
      "learning_rate": 3.575580438718738e-05,
      "loss": 0.0244,
      "step": 10018,
      "task_loss": 0.057412195950746536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017170149832963943,
      "epoch": 9.51,
      "learning_rate": 3.574618472555604e-05,
      "loss": 0.0273,
      "step": 10019,
      "task_loss": 0.11874374747276306
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018801424652338028,
      "epoch": 9.52,
      "learning_rate": 3.573656311182848e-05,
      "loss": 0.0281,
      "step": 10020,
      "task_loss": 0.11174070090055466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013433647342026234,
      "epoch": 9.52,
      "learning_rate": 3.5726939547752536e-05,
      "loss": 0.0126,
      "step": 10021,
      "task_loss": 0.005174616351723671
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010112293064594269,
      "epoch": 9.52,
      "learning_rate": 3.5717314035076355e-05,
      "loss": 0.0095,
      "step": 10022,
      "task_loss": 0.003835149109363556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.041021257638931274,
      "epoch": 9.52,
      "learning_rate": 3.570768657554847e-05,
      "loss": 0.0591,
      "step": 10023,
      "task_loss": 0.22215372323989868
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03894636034965515,
      "epoch": 9.52,
      "learning_rate": 3.569805717091775e-05,
      "loss": 0.0423,
      "step": 10024,
      "task_loss": 0.07290509343147278
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019539494067430496,
      "epoch": 9.52,
      "learning_rate": 3.5688425822933414e-05,
      "loss": 0.0342,
      "step": 10025,
      "task_loss": 0.1657709777355194
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03588699549436569,
      "epoch": 9.52,
      "learning_rate": 3.5678792533345055e-05,
      "loss": 0.0434,
      "step": 10026,
      "task_loss": 0.111113041639328
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.058800891041755676,
      "epoch": 9.52,
      "learning_rate": 3.5669157303902604e-05,
      "loss": 0.0656,
      "step": 10027,
      "task_loss": 0.12722201645374298
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02307158336043358,
      "epoch": 9.52,
      "learning_rate": 3.565952013635635e-05,
      "loss": 0.0256,
      "step": 10028,
      "task_loss": 0.04787431284785271
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012087054550647736,
      "epoch": 9.52,
      "learning_rate": 3.564988103245692e-05,
      "loss": 0.0137,
      "step": 10029,
      "task_loss": 0.028024591505527496
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014830454252660275,
      "epoch": 9.53,
      "learning_rate": 3.564023999395531e-05,
      "loss": 0.0216,
      "step": 10030,
      "task_loss": 0.08278335630893707
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020527433604002,
      "epoch": 9.53,
      "learning_rate": 3.563059702260287e-05,
      "loss": 0.0193,
      "step": 10031,
      "task_loss": 0.00791563093662262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08408955484628677,
      "epoch": 9.53,
      "learning_rate": 3.562095212015128e-05,
      "loss": 0.0893,
      "step": 10032,
      "task_loss": 0.13649523258209229
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0584905743598938,
      "epoch": 9.53,
      "learning_rate": 3.5611305288352576e-05,
      "loss": 0.0576,
      "step": 10033,
      "task_loss": 0.04989843815565109
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09185069054365158,
      "epoch": 9.53,
      "learning_rate": 3.560165652895917e-05,
      "loss": 0.0895,
      "step": 10034,
      "task_loss": 0.0679357573390007
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05496165156364441,
      "epoch": 9.53,
      "learning_rate": 3.5592005843723794e-05,
      "loss": 0.06,
      "step": 10035,
      "task_loss": 0.10506260395050049
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02257445454597473,
      "epoch": 9.53,
      "learning_rate": 3.558235323439955e-05,
      "loss": 0.0208,
      "step": 10036,
      "task_loss": 0.004368903115391731
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03320860490202904,
      "epoch": 9.53,
      "learning_rate": 3.557269870273987e-05,
      "loss": 0.0343,
      "step": 10037,
      "task_loss": 0.04424959793686867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03928327187895775,
      "epoch": 9.53,
      "learning_rate": 3.5563042250498556e-05,
      "loss": 0.0388,
      "step": 10038,
      "task_loss": 0.034501951187849045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06444701552391052,
      "epoch": 9.53,
      "learning_rate": 3.555338387942974e-05,
      "loss": 0.0616,
      "step": 10039,
      "task_loss": 0.03561048209667206
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05409969761967659,
      "epoch": 9.53,
      "learning_rate": 3.5543723591287916e-05,
      "loss": 0.0558,
      "step": 10040,
      "task_loss": 0.07060417532920837
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09123320877552032,
      "epoch": 9.54,
      "learning_rate": 3.5534061387827936e-05,
      "loss": 0.1023,
      "step": 10041,
      "task_loss": 0.202234148979187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014315794222056866,
      "epoch": 9.54,
      "learning_rate": 3.552439727080495e-05,
      "loss": 0.0223,
      "step": 10042,
      "task_loss": 0.09408995509147644
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01862926036119461,
      "epoch": 9.54,
      "learning_rate": 3.5514731241974544e-05,
      "loss": 0.0183,
      "step": 10043,
      "task_loss": 0.015779605135321617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026581861078739166,
      "epoch": 9.54,
      "learning_rate": 3.5505063303092545e-05,
      "loss": 0.0252,
      "step": 10044,
      "task_loss": 0.012902749702334404
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020298294723033905,
      "epoch": 9.54,
      "learning_rate": 3.549539345591521e-05,
      "loss": 0.0195,
      "step": 10045,
      "task_loss": 0.012040220201015472
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09082924574613571,
      "epoch": 9.54,
      "learning_rate": 3.5485721702199104e-05,
      "loss": 0.0905,
      "step": 10046,
      "task_loss": 0.0875604897737503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04488201439380646,
      "epoch": 9.54,
      "learning_rate": 3.547604804370116e-05,
      "loss": 0.0489,
      "step": 10047,
      "task_loss": 0.08512827754020691
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024493440985679626,
      "epoch": 9.54,
      "learning_rate": 3.5466372482178635e-05,
      "loss": 0.0231,
      "step": 10048,
      "task_loss": 0.010883405804634094
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03374480828642845,
      "epoch": 9.54,
      "learning_rate": 3.545669501938913e-05,
      "loss": 0.031,
      "step": 10049,
      "task_loss": 0.005990535020828247
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022520992904901505,
      "epoch": 9.54,
      "learning_rate": 3.544701565709063e-05,
      "loss": 0.0242,
      "step": 10050,
      "task_loss": 0.03893338143825531
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.00832436978816986,
      "epoch": 9.55,
      "learning_rate": 3.54373343970414e-05,
      "loss": 0.0157,
      "step": 10051,
      "task_loss": 0.08178244531154633
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0154949389398098,
      "epoch": 9.55,
      "learning_rate": 3.542765124100014e-05,
      "loss": 0.0143,
      "step": 10052,
      "task_loss": 0.0035653971135616302
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01720263808965683,
      "epoch": 9.55,
      "learning_rate": 3.541796619072579e-05,
      "loss": 0.0175,
      "step": 10053,
      "task_loss": 0.020343858748674393
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12314651161432266,
      "epoch": 9.55,
      "learning_rate": 3.540827924797772e-05,
      "loss": 0.1117,
      "step": 10054,
      "task_loss": 0.00850660353899002
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011486424133181572,
      "epoch": 9.55,
      "learning_rate": 3.5398590414515586e-05,
      "loss": 0.0106,
      "step": 10055,
      "task_loss": 0.002300182357430458
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.006632357835769653,
      "epoch": 9.55,
      "learning_rate": 3.5388899692099433e-05,
      "loss": 0.0235,
      "step": 10056,
      "task_loss": 0.17546282708644867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015405992977321148,
      "epoch": 9.55,
      "learning_rate": 3.537920708248961e-05,
      "loss": 0.0295,
      "step": 10057,
      "task_loss": 0.15601424872875214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06356394290924072,
      "epoch": 9.55,
      "learning_rate": 3.536951258744684e-05,
      "loss": 0.0721,
      "step": 10058,
      "task_loss": 0.1491929441690445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012521881610155106,
      "epoch": 9.55,
      "learning_rate": 3.5359816208732164e-05,
      "loss": 0.0118,
      "step": 10059,
      "task_loss": 0.005132080987095833
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022264419123530388,
      "epoch": 9.55,
      "learning_rate": 3.535011794810698e-05,
      "loss": 0.0216,
      "step": 10060,
      "task_loss": 0.015442634001374245
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022136827930808067,
      "epoch": 9.55,
      "learning_rate": 3.5340417807333026e-05,
      "loss": 0.0227,
      "step": 10061,
      "task_loss": 0.02739689312875271
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013634988106787205,
      "epoch": 9.56,
      "learning_rate": 3.533071578817239e-05,
      "loss": 0.0204,
      "step": 10062,
      "task_loss": 0.08093855530023575
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012208834290504456,
      "epoch": 9.56,
      "learning_rate": 3.5321011892387455e-05,
      "loss": 0.0285,
      "step": 10063,
      "task_loss": 0.1753706932067871
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07617907971143723,
      "epoch": 9.56,
      "learning_rate": 3.5311306121741015e-05,
      "loss": 0.0732,
      "step": 10064,
      "task_loss": 0.046028852462768555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010005553252995014,
      "epoch": 9.56,
      "learning_rate": 3.530159847799616e-05,
      "loss": 0.0097,
      "step": 10065,
      "task_loss": 0.0069500841200351715
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0498456135392189,
      "epoch": 9.56,
      "learning_rate": 3.529188896291632e-05,
      "loss": 0.0624,
      "step": 10066,
      "task_loss": 0.17580851912498474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01079606544226408,
      "epoch": 9.56,
      "learning_rate": 3.5282177578265296e-05,
      "loss": 0.01,
      "step": 10067,
      "task_loss": 0.003003546968102455
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06311812996864319,
      "epoch": 9.56,
      "learning_rate": 3.527246432580718e-05,
      "loss": 0.0622,
      "step": 10068,
      "task_loss": 0.05363881587982178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020096108317375183,
      "epoch": 9.56,
      "learning_rate": 3.526274920730645e-05,
      "loss": 0.0185,
      "step": 10069,
      "task_loss": 0.004027027636766434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016499292105436325,
      "epoch": 9.56,
      "learning_rate": 3.525303222452791e-05,
      "loss": 0.0156,
      "step": 10070,
      "task_loss": 0.007455941289663315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014504550956189632,
      "epoch": 9.56,
      "learning_rate": 3.5243313379236685e-05,
      "loss": 0.0182,
      "step": 10071,
      "task_loss": 0.051125288009643555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028046708554029465,
      "epoch": 9.57,
      "learning_rate": 3.5233592673198245e-05,
      "loss": 0.037,
      "step": 10072,
      "task_loss": 0.11712638288736343
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.048199012875556946,
      "epoch": 9.57,
      "learning_rate": 3.522387010817842e-05,
      "loss": 0.0448,
      "step": 10073,
      "task_loss": 0.013714082539081573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.12250658124685287,
      "epoch": 9.57,
      "learning_rate": 3.521414568594335e-05,
      "loss": 0.1211,
      "step": 10074,
      "task_loss": 0.10803371667861938
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028487298637628555,
      "epoch": 9.57,
      "learning_rate": 3.520441940825952e-05,
      "loss": 0.0274,
      "step": 10075,
      "task_loss": 0.017971640452742577
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01495126262307167,
      "epoch": 9.57,
      "learning_rate": 3.5194691276893755e-05,
      "loss": 0.0234,
      "step": 10076,
      "task_loss": 0.09949477761983871
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018718678504228592,
      "epoch": 9.57,
      "learning_rate": 3.518496129361323e-05,
      "loss": 0.0328,
      "step": 10077,
      "task_loss": 0.1593606173992157
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0187015850096941,
      "epoch": 9.57,
      "learning_rate": 3.5175229460185425e-05,
      "loss": 0.0173,
      "step": 10078,
      "task_loss": 0.004929153248667717
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02536281757056713,
      "epoch": 9.57,
      "learning_rate": 3.5165495778378196e-05,
      "loss": 0.0232,
      "step": 10079,
      "task_loss": 0.00400426983833313
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0379653237760067,
      "epoch": 9.57,
      "learning_rate": 3.51557602499597e-05,
      "loss": 0.0348,
      "step": 10080,
      "task_loss": 0.006340883672237396
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.047561101615428925,
      "epoch": 9.57,
      "learning_rate": 3.514602287669844e-05,
      "loss": 0.0578,
      "step": 10081,
      "task_loss": 0.15020343661308289
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017001332715153694,
      "epoch": 9.57,
      "learning_rate": 3.5136283660363255e-05,
      "loss": 0.0169,
      "step": 10082,
      "task_loss": 0.015564844012260437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025101391598582268,
      "epoch": 9.58,
      "learning_rate": 3.5126542602723334e-05,
      "loss": 0.0319,
      "step": 10083,
      "task_loss": 0.09269939363002777
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.049932364374399185,
      "epoch": 9.58,
      "learning_rate": 3.5116799705548175e-05,
      "loss": 0.0726,
      "step": 10084,
      "task_loss": 0.2766878008842468
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01646530255675316,
      "epoch": 9.58,
      "learning_rate": 3.510705497060762e-05,
      "loss": 0.0153,
      "step": 10085,
      "task_loss": 0.004672164097428322
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.031244024634361267,
      "epoch": 9.58,
      "learning_rate": 3.509730839967187e-05,
      "loss": 0.0304,
      "step": 10086,
      "task_loss": 0.022954711690545082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0413852296769619,
      "epoch": 9.58,
      "learning_rate": 3.508755999451141e-05,
      "loss": 0.0503,
      "step": 10087,
      "task_loss": 0.13052834570407867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024242881685495377,
      "epoch": 9.58,
      "learning_rate": 3.50778097568971e-05,
      "loss": 0.0226,
      "step": 10088,
      "task_loss": 0.008248705416917801
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014527438208460808,
      "epoch": 9.58,
      "learning_rate": 3.506805768860011e-05,
      "loss": 0.0136,
      "step": 10089,
      "task_loss": 0.005055870860815048
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.043073639273643494,
      "epoch": 9.58,
      "learning_rate": 3.505830379139195e-05,
      "loss": 0.0481,
      "step": 10090,
      "task_loss": 0.09338083863258362
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018400005996227264,
      "epoch": 9.58,
      "learning_rate": 3.504854806704446e-05,
      "loss": 0.017,
      "step": 10091,
      "task_loss": 0.004409752786159515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03231433779001236,
      "epoch": 9.58,
      "learning_rate": 3.503879051732983e-05,
      "loss": 0.0422,
      "step": 10092,
      "task_loss": 0.1311528980731964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04105234518647194,
      "epoch": 9.58,
      "learning_rate": 3.502903114402055e-05,
      "loss": 0.0461,
      "step": 10093,
      "task_loss": 0.09118282049894333
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04761913791298866,
      "epoch": 9.59,
      "learning_rate": 3.501926994888946e-05,
      "loss": 0.0462,
      "step": 10094,
      "task_loss": 0.0329703614115715
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013278797268867493,
      "epoch": 9.59,
      "learning_rate": 3.500950693370974e-05,
      "loss": 0.0124,
      "step": 10095,
      "task_loss": 0.004137454554438591
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021566402167081833,
      "epoch": 9.59,
      "learning_rate": 3.499974210025487e-05,
      "loss": 0.0255,
      "step": 10096,
      "task_loss": 0.06101330369710922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03355410322546959,
      "epoch": 9.59,
      "learning_rate": 3.4989975450298694e-05,
      "loss": 0.0436,
      "step": 10097,
      "task_loss": 0.13387075066566467
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011760689318180084,
      "epoch": 9.59,
      "learning_rate": 3.498020698561536e-05,
      "loss": 0.0109,
      "step": 10098,
      "task_loss": 0.0035722479224205017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.040274716913700104,
      "epoch": 9.59,
      "learning_rate": 3.497043670797936e-05,
      "loss": 0.0377,
      "step": 10099,
      "task_loss": 0.014515921473503113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019877292215824127,
      "epoch": 9.59,
      "learning_rate": 3.496066461916552e-05,
      "loss": 0.0187,
      "step": 10100,
      "task_loss": 0.0076863933354616165
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05636334791779518,
      "epoch": 9.59,
      "learning_rate": 3.495089072094898e-05,
      "loss": 0.0547,
      "step": 10101,
      "task_loss": 0.040151506662368774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02113344892859459,
      "epoch": 9.59,
      "learning_rate": 3.494111501510522e-05,
      "loss": 0.0195,
      "step": 10102,
      "task_loss": 0.0052163973450660706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015629485249519348,
      "epoch": 9.59,
      "learning_rate": 3.4931337503410034e-05,
      "loss": 0.0238,
      "step": 10103,
      "task_loss": 0.09734653681516647
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013077488169074059,
      "epoch": 9.6,
      "learning_rate": 3.4921558187639556e-05,
      "loss": 0.0125,
      "step": 10104,
      "task_loss": 0.006894918158650398
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014943759888410568,
      "epoch": 9.6,
      "learning_rate": 3.491177706957026e-05,
      "loss": 0.0193,
      "step": 10105,
      "task_loss": 0.058562587946653366
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05631723254919052,
      "epoch": 9.6,
      "learning_rate": 3.490199415097892e-05,
      "loss": 0.0563,
      "step": 10106,
      "task_loss": 0.056095875799655914
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017733752727508545,
      "epoch": 9.6,
      "learning_rate": 3.489220943364266e-05,
      "loss": 0.0165,
      "step": 10107,
      "task_loss": 0.00526992604136467
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10405229032039642,
      "epoch": 9.6,
      "learning_rate": 3.488242291933891e-05,
      "loss": 0.1191,
      "step": 10108,
      "task_loss": 0.254643976688385
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008884737268090248,
      "epoch": 9.6,
      "learning_rate": 3.487263460984544e-05,
      "loss": 0.0083,
      "step": 10109,
      "task_loss": 0.0033239684998989105
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.00749258091673255,
      "epoch": 9.6,
      "learning_rate": 3.486284450694035e-05,
      "loss": 0.007,
      "step": 10110,
      "task_loss": 0.0027102380990982056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.046791236847639084,
      "epoch": 9.6,
      "learning_rate": 3.485305261240205e-05,
      "loss": 0.0484,
      "step": 10111,
      "task_loss": 0.06307273358106613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026928380131721497,
      "epoch": 9.6,
      "learning_rate": 3.4843258928009294e-05,
      "loss": 0.0259,
      "step": 10112,
      "task_loss": 0.01632116362452507
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030285503715276718,
      "epoch": 9.6,
      "learning_rate": 3.4833463455541146e-05,
      "loss": 0.0344,
      "step": 10113,
      "task_loss": 0.07147219032049179
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027739666402339935,
      "epoch": 9.6,
      "learning_rate": 3.4823666196777006e-05,
      "loss": 0.026,
      "step": 10114,
      "task_loss": 0.01077285036444664
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1198190227150917,
      "epoch": 9.61,
      "learning_rate": 3.481386715349659e-05,
      "loss": 0.119,
      "step": 10115,
      "task_loss": 0.11184495687484741
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02876965142786503,
      "epoch": 9.61,
      "learning_rate": 3.4804066327479934e-05,
      "loss": 0.0307,
      "step": 10116,
      "task_loss": 0.04767701029777527
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06963972747325897,
      "epoch": 9.61,
      "learning_rate": 3.4794263720507427e-05,
      "loss": 0.0748,
      "step": 10117,
      "task_loss": 0.12130016088485718
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10158136487007141,
      "epoch": 9.61,
      "learning_rate": 3.478445933435973e-05,
      "loss": 0.0977,
      "step": 10118,
      "task_loss": 0.06303709745407104
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06906269490718842,
      "epoch": 9.61,
      "learning_rate": 3.4774653170817884e-05,
      "loss": 0.0782,
      "step": 10119,
      "task_loss": 0.1606631875038147
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020416613668203354,
      "epoch": 9.61,
      "learning_rate": 3.4764845231663205e-05,
      "loss": 0.0258,
      "step": 10120,
      "task_loss": 0.07442939281463623
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014320943504571915,
      "epoch": 9.61,
      "learning_rate": 3.475503551867737e-05,
      "loss": 0.0134,
      "step": 10121,
      "task_loss": 0.004721490666270256
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04403253272175789,
      "epoch": 9.61,
      "learning_rate": 3.474522403364235e-05,
      "loss": 0.0434,
      "step": 10122,
      "task_loss": 0.03757096827030182
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009927413426339626,
      "epoch": 9.61,
      "learning_rate": 3.473541077834045e-05,
      "loss": 0.0096,
      "step": 10123,
      "task_loss": 0.0061761606484651566
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.046677879989147186,
      "epoch": 9.61,
      "learning_rate": 3.4725595754554295e-05,
      "loss": 0.0439,
      "step": 10124,
      "task_loss": 0.019169652834534645
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01028907299041748,
      "epoch": 9.62,
      "learning_rate": 3.471577896406683e-05,
      "loss": 0.0096,
      "step": 10125,
      "task_loss": 0.0030922014266252518
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022475387901067734,
      "epoch": 9.62,
      "learning_rate": 3.470596040866133e-05,
      "loss": 0.0296,
      "step": 10126,
      "task_loss": 0.09410066902637482
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010681994259357452,
      "epoch": 9.62,
      "learning_rate": 3.4696140090121376e-05,
      "loss": 0.0101,
      "step": 10127,
      "task_loss": 0.004838347434997559
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011722270399332047,
      "epoch": 9.62,
      "learning_rate": 3.468631801023088e-05,
      "loss": 0.0287,
      "step": 10128,
      "task_loss": 0.18153433501720428
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022453030571341515,
      "epoch": 9.62,
      "learning_rate": 3.467649417077406e-05,
      "loss": 0.0258,
      "step": 10129,
      "task_loss": 0.05557314306497574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08180783689022064,
      "epoch": 9.62,
      "learning_rate": 3.466666857353547e-05,
      "loss": 0.0942,
      "step": 10130,
      "task_loss": 0.20586249232292175
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017414044588804245,
      "epoch": 9.62,
      "learning_rate": 3.465684122029999e-05,
      "loss": 0.0161,
      "step": 10131,
      "task_loss": 0.004198454320430756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.23380213975906372,
      "epoch": 9.62,
      "learning_rate": 3.464701211285279e-05,
      "loss": 0.2391,
      "step": 10132,
      "task_loss": 0.2872486114501953
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013667404651641846,
      "epoch": 9.62,
      "learning_rate": 3.463718125297937e-05,
      "loss": 0.0128,
      "step": 10133,
      "task_loss": 0.004925645887851715
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05431966111063957,
      "epoch": 9.62,
      "learning_rate": 3.462734864246557e-05,
      "loss": 0.0529,
      "step": 10134,
      "task_loss": 0.04007718339562416
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030022673308849335,
      "epoch": 9.62,
      "learning_rate": 3.4617514283097524e-05,
      "loss": 0.0402,
      "step": 10135,
      "task_loss": 0.13180501759052277
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019991595298051834,
      "epoch": 9.63,
      "learning_rate": 3.4607678176661695e-05,
      "loss": 0.0377,
      "step": 10136,
      "task_loss": 0.19743065536022186
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04344207048416138,
      "epoch": 9.63,
      "learning_rate": 3.459784032494484e-05,
      "loss": 0.0485,
      "step": 10137,
      "task_loss": 0.09402826428413391
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.007493377663195133,
      "epoch": 9.63,
      "learning_rate": 3.458800072973408e-05,
      "loss": 0.0071,
      "step": 10138,
      "task_loss": 0.0033276528120040894
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08084356784820557,
      "epoch": 9.63,
      "learning_rate": 3.45781593928168e-05,
      "loss": 0.0823,
      "step": 10139,
      "task_loss": 0.09574754536151886
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01496205572038889,
      "epoch": 9.63,
      "learning_rate": 3.4568316315980745e-05,
      "loss": 0.0204,
      "step": 10140,
      "task_loss": 0.06947852671146393
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024571670219302177,
      "epoch": 9.63,
      "learning_rate": 3.455847150101395e-05,
      "loss": 0.0364,
      "step": 10141,
      "task_loss": 0.14293412864208221
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03470940142869949,
      "epoch": 9.63,
      "learning_rate": 3.454862494970477e-05,
      "loss": 0.0461,
      "step": 10142,
      "task_loss": 0.1487175077199936
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014040854759514332,
      "epoch": 9.63,
      "learning_rate": 3.4538776663841875e-05,
      "loss": 0.0133,
      "step": 10143,
      "task_loss": 0.006179919466376305
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02388843148946762,
      "epoch": 9.63,
      "learning_rate": 3.452892664521427e-05,
      "loss": 0.0248,
      "step": 10144,
      "task_loss": 0.032703425735235214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.007051503751426935,
      "epoch": 9.63,
      "learning_rate": 3.4519074895611244e-05,
      "loss": 0.0067,
      "step": 10145,
      "task_loss": 0.004012970253825188
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013478701002895832,
      "epoch": 9.64,
      "learning_rate": 3.4509221416822415e-05,
      "loss": 0.0126,
      "step": 10146,
      "task_loss": 0.004567300900816917
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.044648945331573486,
      "epoch": 9.64,
      "learning_rate": 3.4499366210637725e-05,
      "loss": 0.0537,
      "step": 10147,
      "task_loss": 0.13514164090156555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017728157341480255,
      "epoch": 9.64,
      "learning_rate": 3.4489509278847414e-05,
      "loss": 0.024,
      "step": 10148,
      "task_loss": 0.08038611710071564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0526009202003479,
      "epoch": 9.64,
      "learning_rate": 3.4479650623242036e-05,
      "loss": 0.0529,
      "step": 10149,
      "task_loss": 0.05552627891302109
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.031424228101968765,
      "epoch": 9.64,
      "learning_rate": 3.446979024561246e-05,
      "loss": 0.0368,
      "step": 10150,
      "task_loss": 0.0853959321975708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08394592255353928,
      "epoch": 9.64,
      "learning_rate": 3.44599281477499e-05,
      "loss": 0.08,
      "step": 10151,
      "task_loss": 0.044918350875377655
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06996019929647446,
      "epoch": 9.64,
      "learning_rate": 3.4450064331445814e-05,
      "loss": 0.0865,
      "step": 10152,
      "task_loss": 0.23547132313251495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.104726642370224,
      "epoch": 9.64,
      "learning_rate": 3.444019879849204e-05,
      "loss": 0.1095,
      "step": 10153,
      "task_loss": 0.15200763940811157
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06347490102052689,
      "epoch": 9.64,
      "learning_rate": 3.443033155068069e-05,
      "loss": 0.0633,
      "step": 10154,
      "task_loss": 0.06222536787390709
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10413339734077454,
      "epoch": 9.64,
      "learning_rate": 3.442046258980419e-05,
      "loss": 0.1025,
      "step": 10155,
      "task_loss": 0.08788105845451355
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04322014003992081,
      "epoch": 9.64,
      "learning_rate": 3.4410591917655296e-05,
      "loss": 0.0468,
      "step": 10156,
      "task_loss": 0.07900369167327881
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.051760174334049225,
      "epoch": 9.65,
      "learning_rate": 3.4400719536027056e-05,
      "loss": 0.0559,
      "step": 10157,
      "task_loss": 0.09334629774093628
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020617887377738953,
      "epoch": 9.65,
      "learning_rate": 3.4390845446712836e-05,
      "loss": 0.0218,
      "step": 10158,
      "task_loss": 0.032847531139850616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.040341269224882126,
      "epoch": 9.65,
      "learning_rate": 3.438096965150632e-05,
      "loss": 0.0382,
      "step": 10159,
      "task_loss": 0.018778573721647263
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01803889125585556,
      "epoch": 9.65,
      "learning_rate": 3.4371092152201485e-05,
      "loss": 0.0195,
      "step": 10160,
      "task_loss": 0.03258706256747246
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05206110328435898,
      "epoch": 9.65,
      "learning_rate": 3.4361212950592624e-05,
      "loss": 0.0492,
      "step": 10161,
      "task_loss": 0.023629793897271156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011896253563463688,
      "epoch": 9.65,
      "learning_rate": 3.435133204847435e-05,
      "loss": 0.0111,
      "step": 10162,
      "task_loss": 0.00429350882768631
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019481834024190903,
      "epoch": 9.65,
      "learning_rate": 3.4341449447641575e-05,
      "loss": 0.0182,
      "step": 10163,
      "task_loss": 0.006555115804076195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.044923312962055206,
      "epoch": 9.65,
      "learning_rate": 3.433156514988951e-05,
      "loss": 0.0486,
      "step": 10164,
      "task_loss": 0.0811932161450386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026373110711574554,
      "epoch": 9.65,
      "learning_rate": 3.432167915701371e-05,
      "loss": 0.0293,
      "step": 10165,
      "task_loss": 0.05567855015397072
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015387404710054398,
      "epoch": 9.65,
      "learning_rate": 3.431179147080999e-05,
      "loss": 0.0244,
      "step": 10166,
      "task_loss": 0.10537419468164444
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05314071476459503,
      "epoch": 9.66,
      "learning_rate": 3.4301902093074504e-05,
      "loss": 0.0508,
      "step": 10167,
      "task_loss": 0.030011305585503578
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020540203899145126,
      "epoch": 9.66,
      "learning_rate": 3.42920110256037e-05,
      "loss": 0.0278,
      "step": 10168,
      "task_loss": 0.09346860647201538
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05021016672253609,
      "epoch": 9.66,
      "learning_rate": 3.428211827019434e-05,
      "loss": 0.0555,
      "step": 10169,
      "task_loss": 0.10355982184410095
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.00913775060325861,
      "epoch": 9.66,
      "learning_rate": 3.42722238286435e-05,
      "loss": 0.0171,
      "step": 10170,
      "task_loss": 0.08899959921836853
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03822232410311699,
      "epoch": 9.66,
      "learning_rate": 3.426232770274855e-05,
      "loss": 0.04,
      "step": 10171,
      "task_loss": 0.05588465929031372
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010778695344924927,
      "epoch": 9.66,
      "learning_rate": 3.4252429894307154e-05,
      "loss": 0.0185,
      "step": 10172,
      "task_loss": 0.08806253969669342
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025916989892721176,
      "epoch": 9.66,
      "learning_rate": 3.424253040511731e-05,
      "loss": 0.0386,
      "step": 10173,
      "task_loss": 0.1526973396539688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015541073866188526,
      "epoch": 9.66,
      "learning_rate": 3.4232629236977316e-05,
      "loss": 0.0146,
      "step": 10174,
      "task_loss": 0.006028560921549797
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0131698502227664,
      "epoch": 9.66,
      "learning_rate": 3.4222726391685746e-05,
      "loss": 0.0269,
      "step": 10175,
      "task_loss": 0.15060535073280334
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01671590656042099,
      "epoch": 9.66,
      "learning_rate": 3.42128218710415e-05,
      "loss": 0.019,
      "step": 10176,
      "task_loss": 0.039641670882701874
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009115570224821568,
      "epoch": 9.66,
      "learning_rate": 3.420291567684381e-05,
      "loss": 0.0088,
      "step": 10177,
      "task_loss": 0.00632680207490921
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08051498234272003,
      "epoch": 9.67,
      "learning_rate": 3.419300781089216e-05,
      "loss": 0.0786,
      "step": 10178,
      "task_loss": 0.061485860496759415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03897204250097275,
      "epoch": 9.67,
      "learning_rate": 3.418309827498637e-05,
      "loss": 0.0359,
      "step": 10179,
      "task_loss": 0.008243357762694359
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011700714007019997,
      "epoch": 9.67,
      "learning_rate": 3.4173187070926546e-05,
      "loss": 0.0111,
      "step": 10180,
      "task_loss": 0.005291668698191643
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012526036240160465,
      "epoch": 9.67,
      "learning_rate": 3.4163274200513116e-05,
      "loss": 0.0186,
      "step": 10181,
      "task_loss": 0.07343530654907227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017924156039953232,
      "epoch": 9.67,
      "learning_rate": 3.415335966554679e-05,
      "loss": 0.0179,
      "step": 10182,
      "task_loss": 0.017576100304722786
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06711853295564651,
      "epoch": 9.67,
      "learning_rate": 3.414344346782861e-05,
      "loss": 0.0728,
      "step": 10183,
      "task_loss": 0.12400171160697937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017617417499423027,
      "epoch": 9.67,
      "learning_rate": 3.413352560915988e-05,
      "loss": 0.0164,
      "step": 10184,
      "task_loss": 0.005933165550231934
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04335098713636398,
      "epoch": 9.67,
      "learning_rate": 3.412360609134223e-05,
      "loss": 0.0492,
      "step": 10185,
      "task_loss": 0.10230584442615509
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01754794456064701,
      "epoch": 9.67,
      "learning_rate": 3.411368491617761e-05,
      "loss": 0.0165,
      "step": 10186,
      "task_loss": 0.006676128134131432
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08341458439826965,
      "epoch": 9.67,
      "learning_rate": 3.410376208546822e-05,
      "loss": 0.0941,
      "step": 10187,
      "task_loss": 0.19037111103534698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012632082216441631,
      "epoch": 9.68,
      "learning_rate": 3.409383760101661e-05,
      "loss": 0.0321,
      "step": 10188,
      "task_loss": 0.20703136920928955
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05429381877183914,
      "epoch": 9.68,
      "learning_rate": 3.4083911464625596e-05,
      "loss": 0.0664,
      "step": 10189,
      "task_loss": 0.17530539631843567
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01827721670269966,
      "epoch": 9.68,
      "learning_rate": 3.407398367809832e-05,
      "loss": 0.0346,
      "step": 10190,
      "task_loss": 0.18168266117572784
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015146693214774132,
      "epoch": 9.68,
      "learning_rate": 3.406405424323821e-05,
      "loss": 0.0274,
      "step": 10191,
      "task_loss": 0.13719764351844788
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05147261545062065,
      "epoch": 9.68,
      "learning_rate": 3.4054123161848995e-05,
      "loss": 0.059,
      "step": 10192,
      "task_loss": 0.12712138891220093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024947544559836388,
      "epoch": 9.68,
      "learning_rate": 3.4044190435734695e-05,
      "loss": 0.0233,
      "step": 10193,
      "task_loss": 0.008586343377828598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010366448201239109,
      "epoch": 9.68,
      "learning_rate": 3.403425606669965e-05,
      "loss": 0.0099,
      "step": 10194,
      "task_loss": 0.005786292254924774
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08639805763959885,
      "epoch": 9.68,
      "learning_rate": 3.4024320056548475e-05,
      "loss": 0.0934,
      "step": 10195,
      "task_loss": 0.15593896806240082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04102171212434769,
      "epoch": 9.68,
      "learning_rate": 3.401438240708611e-05,
      "loss": 0.0455,
      "step": 10196,
      "task_loss": 0.08535470068454742
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02099025435745716,
      "epoch": 9.68,
      "learning_rate": 3.400444312011776e-05,
      "loss": 0.0239,
      "step": 10197,
      "task_loss": 0.049628641456365585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07851079851388931,
      "epoch": 9.68,
      "learning_rate": 3.399450219744896e-05,
      "loss": 0.0751,
      "step": 10198,
      "task_loss": 0.044152356684207916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.039639078080654144,
      "epoch": 9.69,
      "learning_rate": 3.3984559640885505e-05,
      "loss": 0.0609,
      "step": 10199,
      "task_loss": 0.2518640458583832
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03628784418106079,
      "epoch": 9.69,
      "learning_rate": 3.3974615452233526e-05,
      "loss": 0.0336,
      "step": 10200,
      "task_loss": 0.009241720661520958
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07165795564651489,
      "epoch": 9.69,
      "learning_rate": 3.396466963329944e-05,
      "loss": 0.071,
      "step": 10201,
      "task_loss": 0.06518401205539703
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026513490825891495,
      "epoch": 9.69,
      "learning_rate": 3.395472218588992e-05,
      "loss": 0.0297,
      "step": 10202,
      "task_loss": 0.05858299508690834
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023471450433135033,
      "epoch": 9.69,
      "learning_rate": 3.394477311181201e-05,
      "loss": 0.0217,
      "step": 10203,
      "task_loss": 0.005670515820384026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08482164889574051,
      "epoch": 9.69,
      "learning_rate": 3.393482241287297e-05,
      "loss": 0.0974,
      "step": 10204,
      "task_loss": 0.21033701300621033
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05313311517238617,
      "epoch": 9.69,
      "learning_rate": 3.392487009088041e-05,
      "loss": 0.0544,
      "step": 10205,
      "task_loss": 0.06595422327518463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03181048855185509,
      "epoch": 9.69,
      "learning_rate": 3.391491614764222e-05,
      "loss": 0.0369,
      "step": 10206,
      "task_loss": 0.08310259878635406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010201860219240189,
      "epoch": 9.69,
      "learning_rate": 3.390496058496657e-05,
      "loss": 0.018,
      "step": 10207,
      "task_loss": 0.0884314775466919
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02600035071372986,
      "epoch": 9.69,
      "learning_rate": 3.3895003404661955e-05,
      "loss": 0.0239,
      "step": 10208,
      "task_loss": 0.005447834730148315
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04286601021885872,
      "epoch": 9.7,
      "learning_rate": 3.3885044608537125e-05,
      "loss": 0.0517,
      "step": 10209,
      "task_loss": 0.13154464960098267
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0452398955821991,
      "epoch": 9.7,
      "learning_rate": 3.387508419840115e-05,
      "loss": 0.0499,
      "step": 10210,
      "task_loss": 0.0916098803281784
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04269058629870415,
      "epoch": 9.7,
      "learning_rate": 3.386512217606339e-05,
      "loss": 0.0601,
      "step": 10211,
      "task_loss": 0.2167372852563858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06171311438083649,
      "epoch": 9.7,
      "learning_rate": 3.385515854333349e-05,
      "loss": 0.0662,
      "step": 10212,
      "task_loss": 0.10651151090860367
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1391063630580902,
      "epoch": 9.7,
      "learning_rate": 3.38451933020214e-05,
      "loss": 0.135,
      "step": 10213,
      "task_loss": 0.09796366840600967
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05202070623636246,
      "epoch": 9.7,
      "learning_rate": 3.383522645393734e-05,
      "loss": 0.0639,
      "step": 10214,
      "task_loss": 0.17111501097679138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04103817790746689,
      "epoch": 9.7,
      "learning_rate": 3.3825258000891846e-05,
      "loss": 0.0503,
      "step": 10215,
      "task_loss": 0.13386118412017822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026339948177337646,
      "epoch": 9.7,
      "learning_rate": 3.381528794469574e-05,
      "loss": 0.0335,
      "step": 10216,
      "task_loss": 0.09778992086648941
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009569020941853523,
      "epoch": 9.7,
      "learning_rate": 3.380531628716012e-05,
      "loss": 0.0091,
      "step": 10217,
      "task_loss": 0.004733014851808548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05844016373157501,
      "epoch": 9.7,
      "learning_rate": 3.3795343030096384e-05,
      "loss": 0.0596,
      "step": 10218,
      "task_loss": 0.06965689361095428
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.061011701822280884,
      "epoch": 9.7,
      "learning_rate": 3.3785368175316226e-05,
      "loss": 0.0728,
      "step": 10219,
      "task_loss": 0.1784566342830658
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01999977044761181,
      "epoch": 9.71,
      "learning_rate": 3.377539172463164e-05,
      "loss": 0.0248,
      "step": 10220,
      "task_loss": 0.06758248805999756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019030677154660225,
      "epoch": 9.71,
      "learning_rate": 3.376541367985488e-05,
      "loss": 0.0196,
      "step": 10221,
      "task_loss": 0.02473231591284275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03495609387755394,
      "epoch": 9.71,
      "learning_rate": 3.3755434042798506e-05,
      "loss": 0.033,
      "step": 10222,
      "task_loss": 0.01556423120200634
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1749984323978424,
      "epoch": 9.71,
      "learning_rate": 3.374545281527538e-05,
      "loss": 0.1663,
      "step": 10223,
      "task_loss": 0.08834327012300491
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012353798374533653,
      "epoch": 9.71,
      "learning_rate": 3.3735469999098615e-05,
      "loss": 0.0178,
      "step": 10224,
      "task_loss": 0.06693007051944733
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08601576089859009,
      "epoch": 9.71,
      "learning_rate": 3.372548559608166e-05,
      "loss": 0.0848,
      "step": 10225,
      "task_loss": 0.07415040582418442
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03538770601153374,
      "epoch": 9.71,
      "learning_rate": 3.3715499608038234e-05,
      "loss": 0.0404,
      "step": 10226,
      "task_loss": 0.08567916601896286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025410983711481094,
      "epoch": 9.71,
      "learning_rate": 3.370551203678233e-05,
      "loss": 0.0321,
      "step": 10227,
      "task_loss": 0.09199456870555878
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07486201822757721,
      "epoch": 9.71,
      "learning_rate": 3.369552288412822e-05,
      "loss": 0.083,
      "step": 10228,
      "task_loss": 0.15588508546352386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05014657974243164,
      "epoch": 9.71,
      "learning_rate": 3.368553215189052e-05,
      "loss": 0.0607,
      "step": 10229,
      "task_loss": 0.155786395072937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.044199906289577484,
      "epoch": 9.72,
      "learning_rate": 3.367553984188407e-05,
      "loss": 0.0514,
      "step": 10230,
      "task_loss": 0.11591973900794983
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02194777876138687,
      "epoch": 9.72,
      "learning_rate": 3.366554595592402e-05,
      "loss": 0.0213,
      "step": 10231,
      "task_loss": 0.01566997356712818
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06304118037223816,
      "epoch": 9.72,
      "learning_rate": 3.365555049582582e-05,
      "loss": 0.0571,
      "step": 10232,
      "task_loss": 0.003476545214653015
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05096285790205002,
      "epoch": 9.72,
      "learning_rate": 3.364555346340518e-05,
      "loss": 0.0523,
      "step": 10233,
      "task_loss": 0.06410223990678787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.061933599412441254,
      "epoch": 9.72,
      "learning_rate": 3.3635554860478126e-05,
      "loss": 0.0585,
      "step": 10234,
      "task_loss": 0.02772807516157627
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.16535219550132751,
      "epoch": 9.72,
      "learning_rate": 3.362555468886093e-05,
      "loss": 0.1643,
      "step": 10235,
      "task_loss": 0.1545773446559906
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.039442308247089386,
      "epoch": 9.72,
      "learning_rate": 3.361555295037019e-05,
      "loss": 0.047,
      "step": 10236,
      "task_loss": 0.11461587995290756
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019081950187683105,
      "epoch": 9.72,
      "learning_rate": 3.360554964682276e-05,
      "loss": 0.0304,
      "step": 10237,
      "task_loss": 0.13198192417621613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02817378006875515,
      "epoch": 9.72,
      "learning_rate": 3.359554478003579e-05,
      "loss": 0.038,
      "step": 10238,
      "task_loss": 0.1261017769575119
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024465490132570267,
      "epoch": 9.72,
      "learning_rate": 3.358553835182673e-05,
      "loss": 0.0231,
      "step": 10239,
      "task_loss": 0.01032637245953083
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026946038007736206,
      "epoch": 9.72,
      "learning_rate": 3.357553036401326e-05,
      "loss": 0.0259,
      "step": 10240,
      "task_loss": 0.016648059710860252
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10884161293506622,
      "epoch": 9.73,
      "learning_rate": 3.356552081841341e-05,
      "loss": 0.1042,
      "step": 10241,
      "task_loss": 0.062114790081977844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.044160258024930954,
      "epoch": 9.73,
      "learning_rate": 3.355550971684545e-05,
      "loss": 0.0446,
      "step": 10242,
      "task_loss": 0.048725277185440063
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016672126948833466,
      "epoch": 9.73,
      "learning_rate": 3.3545497061127946e-05,
      "loss": 0.0278,
      "step": 10243,
      "task_loss": 0.12836672365665436
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025100041180849075,
      "epoch": 9.73,
      "learning_rate": 3.353548285307975e-05,
      "loss": 0.0276,
      "step": 10244,
      "task_loss": 0.050310954451560974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07603813707828522,
      "epoch": 9.73,
      "learning_rate": 3.352546709451998e-05,
      "loss": 0.0717,
      "step": 10245,
      "task_loss": 0.03274556249380112
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06987538933753967,
      "epoch": 9.73,
      "learning_rate": 3.351544978726805e-05,
      "loss": 0.0762,
      "step": 10246,
      "task_loss": 0.1328875869512558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01046474277973175,
      "epoch": 9.73,
      "learning_rate": 3.350543093314366e-05,
      "loss": 0.0115,
      "step": 10247,
      "task_loss": 0.020672082901000977
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018444234505295753,
      "epoch": 9.73,
      "learning_rate": 3.349541053396678e-05,
      "loss": 0.0198,
      "step": 10248,
      "task_loss": 0.03211362287402153
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13752481341362,
      "epoch": 9.73,
      "learning_rate": 3.348538859155766e-05,
      "loss": 0.1363,
      "step": 10249,
      "task_loss": 0.12482313811779022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024666164070367813,
      "epoch": 9.73,
      "learning_rate": 3.347536510773684e-05,
      "loss": 0.031,
      "step": 10250,
      "task_loss": 0.08787593245506287
    },
    {
      "epoch": 9.73,
      "eval_accuracy": 0.8910550458715596,
      "eval_loss": 0.4798487424850464,
      "eval_runtime": 18.1145,
      "eval_samples_per_second": 48.138,
      "eval_steps_per_second": 6.017,
      "step": 10250
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026560328900814056,
      "epoch": 9.74,
      "learning_rate": 3.346534008432513e-05,
      "loss": 0.0258,
      "step": 10251,
      "task_loss": 0.018506888300180435
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017583798617124557,
      "epoch": 9.74,
      "learning_rate": 3.3455313523143615e-05,
      "loss": 0.0163,
      "step": 10252,
      "task_loss": 0.004295343533158302
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06413343548774719,
      "epoch": 9.74,
      "learning_rate": 3.3445285426013685e-05,
      "loss": 0.0662,
      "step": 10253,
      "task_loss": 0.08491207659244537
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07041002064943314,
      "epoch": 9.74,
      "learning_rate": 3.343525579475698e-05,
      "loss": 0.0762,
      "step": 10254,
      "task_loss": 0.12783107161521912
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021717432886362076,
      "epoch": 9.74,
      "learning_rate": 3.342522463119543e-05,
      "loss": 0.0355,
      "step": 10255,
      "task_loss": 0.15983924269676208
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016216669231653214,
      "epoch": 9.74,
      "learning_rate": 3.341519193715127e-05,
      "loss": 0.0152,
      "step": 10256,
      "task_loss": 0.00632026232779026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03318667411804199,
      "epoch": 9.74,
      "learning_rate": 3.340515771444695e-05,
      "loss": 0.0442,
      "step": 10257,
      "task_loss": 0.14321547746658325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0912448838353157,
      "epoch": 9.74,
      "learning_rate": 3.3395121964905265e-05,
      "loss": 0.0919,
      "step": 10258,
      "task_loss": 0.09797343611717224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09608903527259827,
      "epoch": 9.74,
      "learning_rate": 3.338508469034922e-05,
      "loss": 0.1046,
      "step": 10259,
      "task_loss": 0.18118667602539062
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04384881630539894,
      "epoch": 9.74,
      "learning_rate": 3.3375045892602176e-05,
      "loss": 0.0422,
      "step": 10260,
      "task_loss": 0.02693062275648117
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023890579119324684,
      "epoch": 9.74,
      "learning_rate": 3.3365005573487706e-05,
      "loss": 0.0328,
      "step": 10261,
      "task_loss": 0.11346684396266937
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027102213352918625,
      "epoch": 9.75,
      "learning_rate": 3.335496373482969e-05,
      "loss": 0.0256,
      "step": 10262,
      "task_loss": 0.012507695704698563
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07778482139110565,
      "epoch": 9.75,
      "learning_rate": 3.334492037845227e-05,
      "loss": 0.0737,
      "step": 10263,
      "task_loss": 0.037038303911685944
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0290234312415123,
      "epoch": 9.75,
      "learning_rate": 3.333487550617987e-05,
      "loss": 0.0362,
      "step": 10264,
      "task_loss": 0.10083475708961487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013366115279495716,
      "epoch": 9.75,
      "learning_rate": 3.332482911983721e-05,
      "loss": 0.0137,
      "step": 10265,
      "task_loss": 0.01663101837038994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08686914294958115,
      "epoch": 9.75,
      "learning_rate": 3.331478122124924e-05,
      "loss": 0.0917,
      "step": 10266,
      "task_loss": 0.13518640398979187
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01988716423511505,
      "epoch": 9.75,
      "learning_rate": 3.330473181224121e-05,
      "loss": 0.0226,
      "step": 10267,
      "task_loss": 0.04693231359124184
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03620753064751625,
      "epoch": 9.75,
      "learning_rate": 3.3294680894638655e-05,
      "loss": 0.0332,
      "step": 10268,
      "task_loss": 0.0058042556047439575
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08979254215955734,
      "epoch": 9.75,
      "learning_rate": 3.328462847026736e-05,
      "loss": 0.0962,
      "step": 10269,
      "task_loss": 0.15426769852638245
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030872471630573273,
      "epoch": 9.75,
      "learning_rate": 3.327457454095342e-05,
      "loss": 0.0325,
      "step": 10270,
      "task_loss": 0.04736156761646271
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010424168780446053,
      "epoch": 9.75,
      "learning_rate": 3.3264519108523154e-05,
      "loss": 0.0103,
      "step": 10271,
      "task_loss": 0.009468691423535347
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03220530226826668,
      "epoch": 9.75,
      "learning_rate": 3.3254462174803186e-05,
      "loss": 0.0311,
      "step": 10272,
      "task_loss": 0.021126460283994675
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03830189257860184,
      "epoch": 9.76,
      "learning_rate": 3.324440374162041e-05,
      "loss": 0.0357,
      "step": 10273,
      "task_loss": 0.012562312185764313
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03549380600452423,
      "epoch": 9.76,
      "learning_rate": 3.323434381080199e-05,
      "loss": 0.0339,
      "step": 10274,
      "task_loss": 0.019155368208885193
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0406593419611454,
      "epoch": 9.76,
      "learning_rate": 3.322428238417537e-05,
      "loss": 0.0548,
      "step": 10275,
      "task_loss": 0.18189404904842377
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08273656666278839,
      "epoch": 9.76,
      "learning_rate": 3.321421946356823e-05,
      "loss": 0.0821,
      "step": 10276,
      "task_loss": 0.07649406045675278
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.047196559607982635,
      "epoch": 9.76,
      "learning_rate": 3.320415505080858e-05,
      "loss": 0.0473,
      "step": 10277,
      "task_loss": 0.048162683844566345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0971732959151268,
      "epoch": 9.76,
      "learning_rate": 3.3194089147724644e-05,
      "loss": 0.1024,
      "step": 10278,
      "task_loss": 0.1492057889699936
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008488696068525314,
      "epoch": 9.76,
      "learning_rate": 3.3184021756144954e-05,
      "loss": 0.0085,
      "step": 10279,
      "task_loss": 0.008146345615386963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03513052687048912,
      "epoch": 9.76,
      "learning_rate": 3.317395287789829e-05,
      "loss": 0.0421,
      "step": 10280,
      "task_loss": 0.1049308255314827
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.060586679726839066,
      "epoch": 9.76,
      "learning_rate": 3.316388251481373e-05,
      "loss": 0.0639,
      "step": 10281,
      "task_loss": 0.09365570545196533
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02429119497537613,
      "epoch": 9.76,
      "learning_rate": 3.3153810668720594e-05,
      "loss": 0.0301,
      "step": 10282,
      "task_loss": 0.0827622190117836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02623087167739868,
      "epoch": 9.77,
      "learning_rate": 3.3143737341448475e-05,
      "loss": 0.0325,
      "step": 10283,
      "task_loss": 0.0885535329580307
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09432581067085266,
      "epoch": 9.77,
      "learning_rate": 3.3133662534827255e-05,
      "loss": 0.0897,
      "step": 10284,
      "task_loss": 0.04778246581554413
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.040426380932331085,
      "epoch": 9.77,
      "learning_rate": 3.3123586250687055e-05,
      "loss": 0.0396,
      "step": 10285,
      "task_loss": 0.031931400299072266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.047953128814697266,
      "epoch": 9.77,
      "learning_rate": 3.311350849085829e-05,
      "loss": 0.0458,
      "step": 10286,
      "task_loss": 0.026047490537166595
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.041989296674728394,
      "epoch": 9.77,
      "learning_rate": 3.3103429257171635e-05,
      "loss": 0.0387,
      "step": 10287,
      "task_loss": 0.009208640083670616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023370787501335144,
      "epoch": 9.77,
      "learning_rate": 3.309334855145803e-05,
      "loss": 0.031,
      "step": 10288,
      "task_loss": 0.09938006848096848
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07616433501243591,
      "epoch": 9.77,
      "learning_rate": 3.3083266375548675e-05,
      "loss": 0.0742,
      "step": 10289,
      "task_loss": 0.05699377879500389
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13219673931598663,
      "epoch": 9.77,
      "learning_rate": 3.3073182731275064e-05,
      "loss": 0.1273,
      "step": 10290,
      "task_loss": 0.08289719372987747
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.043332893401384354,
      "epoch": 9.77,
      "learning_rate": 3.306309762046892e-05,
      "loss": 0.0515,
      "step": 10291,
      "task_loss": 0.1250426173210144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10248932242393494,
      "epoch": 9.77,
      "learning_rate": 3.305301104496227e-05,
      "loss": 0.0988,
      "step": 10292,
      "task_loss": 0.06527666747570038
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01327238418161869,
      "epoch": 9.77,
      "learning_rate": 3.3042923006587366e-05,
      "loss": 0.0264,
      "step": 10293,
      "task_loss": 0.145028218626976
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015359059907495975,
      "epoch": 9.78,
      "learning_rate": 3.303283350717678e-05,
      "loss": 0.0216,
      "step": 10294,
      "task_loss": 0.07747124135494232
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019833415746688843,
      "epoch": 9.78,
      "learning_rate": 3.302274254856329e-05,
      "loss": 0.0288,
      "step": 10295,
      "task_loss": 0.10919887572526932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015060881152749062,
      "epoch": 9.78,
      "learning_rate": 3.301265013257998e-05,
      "loss": 0.0139,
      "step": 10296,
      "task_loss": 0.0035025514662265778
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13087275624275208,
      "epoch": 9.78,
      "learning_rate": 3.300255626106019e-05,
      "loss": 0.1465,
      "step": 10297,
      "task_loss": 0.2874350845813751
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02529519982635975,
      "epoch": 9.78,
      "learning_rate": 3.2992460935837505e-05,
      "loss": 0.0235,
      "step": 10298,
      "task_loss": 0.0074701253324747086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011823216453194618,
      "epoch": 9.78,
      "learning_rate": 3.2982364158745805e-05,
      "loss": 0.011,
      "step": 10299,
      "task_loss": 0.003194596618413925
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05281532555818558,
      "epoch": 9.78,
      "learning_rate": 3.297226593161921e-05,
      "loss": 0.0509,
      "step": 10300,
      "task_loss": 0.033622805029153824
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013377421535551548,
      "epoch": 9.78,
      "learning_rate": 3.2962166256292113e-05,
      "loss": 0.0233,
      "step": 10301,
      "task_loss": 0.11291710287332535
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011167014017701149,
      "epoch": 9.78,
      "learning_rate": 3.295206513459917e-05,
      "loss": 0.0191,
      "step": 10302,
      "task_loss": 0.09001649916172028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019459430128335953,
      "epoch": 9.78,
      "learning_rate": 3.2941962568375296e-05,
      "loss": 0.018,
      "step": 10303,
      "task_loss": 0.005041791126132011
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014631148427724838,
      "epoch": 9.79,
      "learning_rate": 3.2931858559455674e-05,
      "loss": 0.0183,
      "step": 10304,
      "task_loss": 0.05133789777755737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08178240060806274,
      "epoch": 9.79,
      "learning_rate": 3.292175310967575e-05,
      "loss": 0.085,
      "step": 10305,
      "task_loss": 0.11350104212760925
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029704544693231583,
      "epoch": 9.79,
      "learning_rate": 3.291164622087122e-05,
      "loss": 0.039,
      "step": 10306,
      "task_loss": 0.12229295074939728
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.007734452374279499,
      "epoch": 9.79,
      "learning_rate": 3.290153789487804e-05,
      "loss": 0.0074,
      "step": 10307,
      "task_loss": 0.004266131669282913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01227061077952385,
      "epoch": 9.79,
      "learning_rate": 3.289142813353246e-05,
      "loss": 0.0113,
      "step": 10308,
      "task_loss": 0.0023228712379932404
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016713295131921768,
      "epoch": 9.79,
      "learning_rate": 3.2881316938670945e-05,
      "loss": 0.0228,
      "step": 10309,
      "task_loss": 0.07756958901882172
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.031723689287900925,
      "epoch": 9.79,
      "learning_rate": 3.2871204312130254e-05,
      "loss": 0.032,
      "step": 10310,
      "task_loss": 0.03421415388584137
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015944186598062515,
      "epoch": 9.79,
      "learning_rate": 3.28610902557474e-05,
      "loss": 0.0221,
      "step": 10311,
      "task_loss": 0.07747285813093185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04275937378406525,
      "epoch": 9.79,
      "learning_rate": 3.285097477135963e-05,
      "loss": 0.0503,
      "step": 10312,
      "task_loss": 0.11856701970100403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023751258850097656,
      "epoch": 9.79,
      "learning_rate": 3.284085786080449e-05,
      "loss": 0.023,
      "step": 10313,
      "task_loss": 0.01643642783164978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019033825024962425,
      "epoch": 9.79,
      "learning_rate": 3.283073952591975e-05,
      "loss": 0.0185,
      "step": 10314,
      "task_loss": 0.013599434867501259
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.13899603486061096,
      "epoch": 9.8,
      "learning_rate": 3.2820619768543473e-05,
      "loss": 0.1415,
      "step": 10315,
      "task_loss": 0.1637457311153412
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09794485569000244,
      "epoch": 9.8,
      "learning_rate": 3.281049859051394e-05,
      "loss": 0.0938,
      "step": 10316,
      "task_loss": 0.0560276135802269
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018764419481158257,
      "epoch": 9.8,
      "learning_rate": 3.280037599366972e-05,
      "loss": 0.0268,
      "step": 10317,
      "task_loss": 0.09955213218927383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11311816424131393,
      "epoch": 9.8,
      "learning_rate": 3.2790251979849654e-05,
      "loss": 0.1127,
      "step": 10318,
      "task_loss": 0.10870064795017242
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10437528789043427,
      "epoch": 9.8,
      "learning_rate": 3.278012655089277e-05,
      "loss": 0.126,
      "step": 10319,
      "task_loss": 0.3202948570251465
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009548291563987732,
      "epoch": 9.8,
      "learning_rate": 3.276999970863845e-05,
      "loss": 0.0129,
      "step": 10320,
      "task_loss": 0.04352171719074249
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01687040366232395,
      "epoch": 9.8,
      "learning_rate": 3.275987145492625e-05,
      "loss": 0.0159,
      "step": 10321,
      "task_loss": 0.007143234834074974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.047225743532180786,
      "epoch": 9.8,
      "learning_rate": 3.274974179159603e-05,
      "loss": 0.0458,
      "step": 10322,
      "task_loss": 0.0325019434094429
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03787485137581825,
      "epoch": 9.8,
      "learning_rate": 3.27396107204879e-05,
      "loss": 0.0505,
      "step": 10323,
      "task_loss": 0.16443775594234467
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024566274136304855,
      "epoch": 9.8,
      "learning_rate": 3.2729478243442194e-05,
      "loss": 0.0231,
      "step": 10324,
      "task_loss": 0.009893251582980156
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018858332186937332,
      "epoch": 9.81,
      "learning_rate": 3.271934436229955e-05,
      "loss": 0.0259,
      "step": 10325,
      "task_loss": 0.0887862890958786
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012717785313725471,
      "epoch": 9.81,
      "learning_rate": 3.270920907890082e-05,
      "loss": 0.0124,
      "step": 10326,
      "task_loss": 0.00914292223751545
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008497057482600212,
      "epoch": 9.81,
      "learning_rate": 3.269907239508714e-05,
      "loss": 0.0103,
      "step": 10327,
      "task_loss": 0.026958035305142403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013917407020926476,
      "epoch": 9.81,
      "learning_rate": 3.268893431269987e-05,
      "loss": 0.0173,
      "step": 10328,
      "task_loss": 0.047798462212085724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022593341767787933,
      "epoch": 9.81,
      "learning_rate": 3.2678794833580654e-05,
      "loss": 0.0267,
      "step": 10329,
      "task_loss": 0.0638725757598877
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020902365446090698,
      "epoch": 9.81,
      "learning_rate": 3.2668653959571384e-05,
      "loss": 0.0272,
      "step": 10330,
      "task_loss": 0.08406180143356323
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021388908848166466,
      "epoch": 9.81,
      "learning_rate": 3.2658511692514184e-05,
      "loss": 0.0294,
      "step": 10331,
      "task_loss": 0.10142026096582413
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03885917365550995,
      "epoch": 9.81,
      "learning_rate": 3.2648368034251454e-05,
      "loss": 0.0584,
      "step": 10332,
      "task_loss": 0.23429960012435913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06736599653959274,
      "epoch": 9.81,
      "learning_rate": 3.263822298662583e-05,
      "loss": 0.0732,
      "step": 10333,
      "task_loss": 0.12564432621002197
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01693696156144142,
      "epoch": 9.81,
      "learning_rate": 3.2628076551480216e-05,
      "loss": 0.0215,
      "step": 10334,
      "task_loss": 0.06282781064510345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03203189745545387,
      "epoch": 9.81,
      "learning_rate": 3.2617928730657764e-05,
      "loss": 0.0306,
      "step": 10335,
      "task_loss": 0.018122676759958267
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01440800167620182,
      "epoch": 9.82,
      "learning_rate": 3.260777952600186e-05,
      "loss": 0.0135,
      "step": 10336,
      "task_loss": 0.004851058125495911
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04675164446234703,
      "epoch": 9.82,
      "learning_rate": 3.2597628939356175e-05,
      "loss": 0.0745,
      "step": 10337,
      "task_loss": 0.3238287568092346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012276804074645042,
      "epoch": 9.82,
      "learning_rate": 3.25874769725646e-05,
      "loss": 0.0225,
      "step": 10338,
      "task_loss": 0.114773690700531
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013100717216730118,
      "epoch": 9.82,
      "learning_rate": 3.257732362747129e-05,
      "loss": 0.0128,
      "step": 10339,
      "task_loss": 0.009668847545981407
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07656170427799225,
      "epoch": 9.82,
      "learning_rate": 3.256716890592065e-05,
      "loss": 0.0814,
      "step": 10340,
      "task_loss": 0.1245703175663948
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09409487247467041,
      "epoch": 9.82,
      "learning_rate": 3.255701280975733e-05,
      "loss": 0.0905,
      "step": 10341,
      "task_loss": 0.05802832543849945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.00820472463965416,
      "epoch": 9.82,
      "learning_rate": 3.2546855340826246e-05,
      "loss": 0.0078,
      "step": 10342,
      "task_loss": 0.004141604527831078
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0564790740609169,
      "epoch": 9.82,
      "learning_rate": 3.253669650097254e-05,
      "loss": 0.0648,
      "step": 10343,
      "task_loss": 0.13937750458717346
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.033709198236465454,
      "epoch": 9.82,
      "learning_rate": 3.2526536292041625e-05,
      "loss": 0.0378,
      "step": 10344,
      "task_loss": 0.07450772076845169
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.061526041477918625,
      "epoch": 9.82,
      "learning_rate": 3.2516374715879126e-05,
      "loss": 0.0774,
      "step": 10345,
      "task_loss": 0.22075936198234558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029069392010569572,
      "epoch": 9.83,
      "learning_rate": 3.250621177433097e-05,
      "loss": 0.0391,
      "step": 10346,
      "task_loss": 0.12960076332092285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03016573004424572,
      "epoch": 9.83,
      "learning_rate": 3.249604746924331e-05,
      "loss": 0.0284,
      "step": 10347,
      "task_loss": 0.012552225962281227
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1340094357728958,
      "epoch": 9.83,
      "learning_rate": 3.248588180246251e-05,
      "loss": 0.1435,
      "step": 10348,
      "task_loss": 0.22935834527015686
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.046159401535987854,
      "epoch": 9.83,
      "learning_rate": 3.247571477583523e-05,
      "loss": 0.0571,
      "step": 10349,
      "task_loss": 0.1553247720003128
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01781924068927765,
      "epoch": 9.83,
      "learning_rate": 3.2465546391208355e-05,
      "loss": 0.0167,
      "step": 10350,
      "task_loss": 0.007019467651844025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01618257164955139,
      "epoch": 9.83,
      "learning_rate": 3.245537665042903e-05,
      "loss": 0.0214,
      "step": 10351,
      "task_loss": 0.0686056837439537
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03626525402069092,
      "epoch": 9.83,
      "learning_rate": 3.244520555534463e-05,
      "loss": 0.034,
      "step": 10352,
      "task_loss": 0.014076223596930504
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10587593913078308,
      "epoch": 9.83,
      "learning_rate": 3.243503310780278e-05,
      "loss": 0.102,
      "step": 10353,
      "task_loss": 0.06675413995981216
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014715575613081455,
      "epoch": 9.83,
      "learning_rate": 3.242485930965136e-05,
      "loss": 0.0137,
      "step": 10354,
      "task_loss": 0.004238050431013107
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0231354720890522,
      "epoch": 9.83,
      "learning_rate": 3.241468416273849e-05,
      "loss": 0.0226,
      "step": 10355,
      "task_loss": 0.01763581857085228
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09283407032489777,
      "epoch": 9.83,
      "learning_rate": 3.2404507668912534e-05,
      "loss": 0.0874,
      "step": 10356,
      "task_loss": 0.03887883201241493
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04399856925010681,
      "epoch": 9.84,
      "learning_rate": 3.2394329830022095e-05,
      "loss": 0.0429,
      "step": 10357,
      "task_loss": 0.0327279269695282
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.042265597730875015,
      "epoch": 9.84,
      "learning_rate": 3.238415064791603e-05,
      "loss": 0.0397,
      "step": 10358,
      "task_loss": 0.016846617683768272
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024321310222148895,
      "epoch": 9.84,
      "learning_rate": 3.237397012444344e-05,
      "loss": 0.0254,
      "step": 10359,
      "task_loss": 0.03516604006290436
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03976666182279587,
      "epoch": 9.84,
      "learning_rate": 3.2363788261453664e-05,
      "loss": 0.0445,
      "step": 10360,
      "task_loss": 0.08759070187807083
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07277016341686249,
      "epoch": 9.84,
      "learning_rate": 3.2353605060796286e-05,
      "loss": 0.067,
      "step": 10361,
      "task_loss": 0.015487806871533394
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013543391600251198,
      "epoch": 9.84,
      "learning_rate": 3.2343420524321134e-05,
      "loss": 0.0139,
      "step": 10362,
      "task_loss": 0.01754959300160408
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06739618629217148,
      "epoch": 9.84,
      "learning_rate": 3.2333234653878275e-05,
      "loss": 0.065,
      "step": 10363,
      "task_loss": 0.04387857764959335
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06469708681106567,
      "epoch": 9.84,
      "learning_rate": 3.2323047451318023e-05,
      "loss": 0.0679,
      "step": 10364,
      "task_loss": 0.09665581583976746
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025472253561019897,
      "epoch": 9.84,
      "learning_rate": 3.2312858918490936e-05,
      "loss": 0.0235,
      "step": 10365,
      "task_loss": 0.005352867767214775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08233923465013504,
      "epoch": 9.84,
      "learning_rate": 3.2302669057247806e-05,
      "loss": 0.0887,
      "step": 10366,
      "task_loss": 0.14634165167808533
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09745900332927704,
      "epoch": 9.85,
      "learning_rate": 3.2292477869439666e-05,
      "loss": 0.1002,
      "step": 10367,
      "task_loss": 0.1244800016283989
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0947013646364212,
      "epoch": 9.85,
      "learning_rate": 3.228228535691781e-05,
      "loss": 0.0893,
      "step": 10368,
      "task_loss": 0.04106832295656204
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019390946254134178,
      "epoch": 9.85,
      "learning_rate": 3.2272091521533745e-05,
      "loss": 0.0182,
      "step": 10369,
      "task_loss": 0.007950548082590103
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020949633792042732,
      "epoch": 9.85,
      "learning_rate": 3.226189636513923e-05,
      "loss": 0.0202,
      "step": 10370,
      "task_loss": 0.013539431616663933
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015328247100114822,
      "epoch": 9.85,
      "learning_rate": 3.225169988958627e-05,
      "loss": 0.0147,
      "step": 10371,
      "task_loss": 0.009431937709450722
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029308216646313667,
      "epoch": 9.85,
      "learning_rate": 3.2241502096727095e-05,
      "loss": 0.0282,
      "step": 10372,
      "task_loss": 0.018634788691997528
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.056150779128074646,
      "epoch": 9.85,
      "learning_rate": 3.2231302988414194e-05,
      "loss": 0.0586,
      "step": 10373,
      "task_loss": 0.08058308064937592
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03458091989159584,
      "epoch": 9.85,
      "learning_rate": 3.222110256650028e-05,
      "loss": 0.0484,
      "step": 10374,
      "task_loss": 0.17240062355995178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06620018929243088,
      "epoch": 9.85,
      "learning_rate": 3.2210900832838295e-05,
      "loss": 0.0788,
      "step": 10375,
      "task_loss": 0.1921333372592926
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04633745551109314,
      "epoch": 9.85,
      "learning_rate": 3.220069778928146e-05,
      "loss": 0.0467,
      "step": 10376,
      "task_loss": 0.04954282194375992
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027524154633283615,
      "epoch": 9.85,
      "learning_rate": 3.2190493437683185e-05,
      "loss": 0.0322,
      "step": 10377,
      "task_loss": 0.07475399971008301
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.062353409826755524,
      "epoch": 9.86,
      "learning_rate": 3.2180287779897155e-05,
      "loss": 0.0616,
      "step": 10378,
      "task_loss": 0.05466865003108978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0365450382232666,
      "epoch": 9.86,
      "learning_rate": 3.217008081777726e-05,
      "loss": 0.0391,
      "step": 10379,
      "task_loss": 0.06168461591005325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07208600640296936,
      "epoch": 9.86,
      "learning_rate": 3.2159872553177655e-05,
      "loss": 0.0865,
      "step": 10380,
      "task_loss": 0.21665066480636597
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013081874698400497,
      "epoch": 9.86,
      "learning_rate": 3.2149662987952725e-05,
      "loss": 0.0183,
      "step": 10381,
      "task_loss": 0.0652560293674469
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11617452651262283,
      "epoch": 9.86,
      "learning_rate": 3.213945212395707e-05,
      "loss": 0.112,
      "step": 10382,
      "task_loss": 0.07452362775802612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02089625969529152,
      "epoch": 9.86,
      "learning_rate": 3.212923996304556e-05,
      "loss": 0.0311,
      "step": 10383,
      "task_loss": 0.12253687530755997
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0406123585999012,
      "epoch": 9.86,
      "learning_rate": 3.211902650707327e-05,
      "loss": 0.0457,
      "step": 10384,
      "task_loss": 0.09141912311315536
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08201208710670471,
      "epoch": 9.86,
      "learning_rate": 3.210881175789553e-05,
      "loss": 0.092,
      "step": 10385,
      "task_loss": 0.18144634366035461
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02090279385447502,
      "epoch": 9.86,
      "learning_rate": 3.209859571736791e-05,
      "loss": 0.0203,
      "step": 10386,
      "task_loss": 0.014566395431756973
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05922047048807144,
      "epoch": 9.86,
      "learning_rate": 3.208837838734618e-05,
      "loss": 0.061,
      "step": 10387,
      "task_loss": 0.07677384465932846
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05778921768069267,
      "epoch": 9.87,
      "learning_rate": 3.207815976968638e-05,
      "loss": 0.0572,
      "step": 10388,
      "task_loss": 0.052189189940690994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03705509752035141,
      "epoch": 9.87,
      "learning_rate": 3.2067939866244764e-05,
      "loss": 0.0341,
      "step": 10389,
      "task_loss": 0.00781635195016861
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022746965289115906,
      "epoch": 9.87,
      "learning_rate": 3.205771867887784e-05,
      "loss": 0.0218,
      "step": 10390,
      "task_loss": 0.013284927234053612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04225531220436096,
      "epoch": 9.87,
      "learning_rate": 3.204749620944232e-05,
      "loss": 0.0421,
      "step": 10391,
      "task_loss": 0.041095249354839325
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08849678933620453,
      "epoch": 9.87,
      "learning_rate": 3.203727245979518e-05,
      "loss": 0.0892,
      "step": 10392,
      "task_loss": 0.09528757631778717
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1014036163687706,
      "epoch": 9.87,
      "learning_rate": 3.20270474317936e-05,
      "loss": 0.1055,
      "step": 10393,
      "task_loss": 0.14196741580963135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04590814188122749,
      "epoch": 9.87,
      "learning_rate": 3.201682112729502e-05,
      "loss": 0.0474,
      "step": 10394,
      "task_loss": 0.06046411767601967
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03985198959708214,
      "epoch": 9.87,
      "learning_rate": 3.2006593548157074e-05,
      "loss": 0.0387,
      "step": 10395,
      "task_loss": 0.028516611084342003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06808362156152725,
      "epoch": 9.87,
      "learning_rate": 3.1996364696237676e-05,
      "loss": 0.0643,
      "step": 10396,
      "task_loss": 0.030522214248776436
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03122507780790329,
      "epoch": 9.87,
      "learning_rate": 3.198613457339493e-05,
      "loss": 0.0428,
      "step": 10397,
      "task_loss": 0.14679580926895142
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02945810928940773,
      "epoch": 9.87,
      "learning_rate": 3.19759031814872e-05,
      "loss": 0.0363,
      "step": 10398,
      "task_loss": 0.09743687510490417
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02599484845995903,
      "epoch": 9.88,
      "learning_rate": 3.196567052237306e-05,
      "loss": 0.0254,
      "step": 10399,
      "task_loss": 0.020352628082036972
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.22350770235061646,
      "epoch": 9.88,
      "learning_rate": 3.195543659791132e-05,
      "loss": 0.2232,
      "step": 10400,
      "task_loss": 0.2200383096933365
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.041732676327228546,
      "epoch": 9.88,
      "learning_rate": 3.194520140996102e-05,
      "loss": 0.0469,
      "step": 10401,
      "task_loss": 0.09368692338466644
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.041806213557720184,
      "epoch": 9.88,
      "learning_rate": 3.193496496038144e-05,
      "loss": 0.0622,
      "step": 10402,
      "task_loss": 0.24602389335632324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0394703671336174,
      "epoch": 9.88,
      "learning_rate": 3.1924727251032075e-05,
      "loss": 0.0432,
      "step": 10403,
      "task_loss": 0.07703036814928055
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.034872762858867645,
      "epoch": 9.88,
      "learning_rate": 3.191448828377267e-05,
      "loss": 0.0486,
      "step": 10404,
      "task_loss": 0.17233772575855255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10517394542694092,
      "epoch": 9.88,
      "learning_rate": 3.1904248060463146e-05,
      "loss": 0.1068,
      "step": 10405,
      "task_loss": 0.12141425162553787
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010680675506591797,
      "epoch": 9.88,
      "learning_rate": 3.189400658296372e-05,
      "loss": 0.0104,
      "step": 10406,
      "task_loss": 0.007806859910488129
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09319774061441422,
      "epoch": 9.88,
      "learning_rate": 3.188376385313479e-05,
      "loss": 0.0928,
      "step": 10407,
      "task_loss": 0.08945365995168686
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05169279873371124,
      "epoch": 9.88,
      "learning_rate": 3.187351987283701e-05,
      "loss": 0.0558,
      "step": 10408,
      "task_loss": 0.09317981451749802
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02045218087732792,
      "epoch": 9.89,
      "learning_rate": 3.1863274643931244e-05,
      "loss": 0.0263,
      "step": 10409,
      "task_loss": 0.0785207599401474
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07327836751937866,
      "epoch": 9.89,
      "learning_rate": 3.185302816827858e-05,
      "loss": 0.0837,
      "step": 10410,
      "task_loss": 0.17741957306861877
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012392308562994003,
      "epoch": 9.89,
      "learning_rate": 3.184278044774035e-05,
      "loss": 0.0115,
      "step": 10411,
      "task_loss": 0.003290366381406784
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.052759043872356415,
      "epoch": 9.89,
      "learning_rate": 3.183253148417808e-05,
      "loss": 0.0562,
      "step": 10412,
      "task_loss": 0.08686379343271255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02482384815812111,
      "epoch": 9.89,
      "learning_rate": 3.182228127945358e-05,
      "loss": 0.0357,
      "step": 10413,
      "task_loss": 0.13362696766853333
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03491012752056122,
      "epoch": 9.89,
      "learning_rate": 3.1812029835428825e-05,
      "loss": 0.0356,
      "step": 10414,
      "task_loss": 0.04205203801393509
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06706416606903076,
      "epoch": 9.89,
      "learning_rate": 3.1801777153966034e-05,
      "loss": 0.071,
      "step": 10415,
      "task_loss": 0.1061612218618393
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0414917916059494,
      "epoch": 9.89,
      "learning_rate": 3.179152323692767e-05,
      "loss": 0.0383,
      "step": 10416,
      "task_loss": 0.009932199493050575
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022580554708838463,
      "epoch": 9.89,
      "learning_rate": 3.1781268086176406e-05,
      "loss": 0.0401,
      "step": 10417,
      "task_loss": 0.1979600191116333
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07729755342006683,
      "epoch": 9.89,
      "learning_rate": 3.177101170357513e-05,
      "loss": 0.0813,
      "step": 10418,
      "task_loss": 0.11744911223649979
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017039090394973755,
      "epoch": 9.89,
      "learning_rate": 3.1760754090986975e-05,
      "loss": 0.0158,
      "step": 10419,
      "task_loss": 0.004911573603749275
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.034923627972602844,
      "epoch": 9.9,
      "learning_rate": 3.175049525027527e-05,
      "loss": 0.0393,
      "step": 10420,
      "task_loss": 0.07879183441400528
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015226260758936405,
      "epoch": 9.9,
      "learning_rate": 3.17402351833036e-05,
      "loss": 0.0209,
      "step": 10421,
      "task_loss": 0.07186593115329742
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04073633253574371,
      "epoch": 9.9,
      "learning_rate": 3.1729973891935745e-05,
      "loss": 0.0493,
      "step": 10422,
      "task_loss": 0.12683020532131195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.059145376086235046,
      "epoch": 9.9,
      "learning_rate": 3.1719711378035714e-05,
      "loss": 0.0822,
      "step": 10423,
      "task_loss": 0.2894851565361023
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04453141614794731,
      "epoch": 9.9,
      "learning_rate": 3.1709447643467755e-05,
      "loss": 0.0422,
      "step": 10424,
      "task_loss": 0.021708250045776367
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027283282950520515,
      "epoch": 9.9,
      "learning_rate": 3.1699182690096316e-05,
      "loss": 0.036,
      "step": 10425,
      "task_loss": 0.11464565247297287
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.24350899457931519,
      "epoch": 9.9,
      "learning_rate": 3.168891651978609e-05,
      "loss": 0.2368,
      "step": 10426,
      "task_loss": 0.17603522539138794
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07290820777416229,
      "epoch": 9.9,
      "learning_rate": 3.167864913440195e-05,
      "loss": 0.0691,
      "step": 10427,
      "task_loss": 0.034495919942855835
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05629532411694527,
      "epoch": 9.9,
      "learning_rate": 3.1668380535809036e-05,
      "loss": 0.0606,
      "step": 10428,
      "task_loss": 0.09951600432395935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07032791525125504,
      "epoch": 9.9,
      "learning_rate": 3.165811072587267e-05,
      "loss": 0.0833,
      "step": 10429,
      "task_loss": 0.20026040077209473
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02635234221816063,
      "epoch": 9.91,
      "learning_rate": 3.164783970645845e-05,
      "loss": 0.0334,
      "step": 10430,
      "task_loss": 0.09689018875360489
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014292672276496887,
      "epoch": 9.91,
      "learning_rate": 3.1637567479432113e-05,
      "loss": 0.0284,
      "step": 10431,
      "task_loss": 0.15557925403118134
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.038801245391368866,
      "epoch": 9.91,
      "learning_rate": 3.1627294046659675e-05,
      "loss": 0.042,
      "step": 10432,
      "task_loss": 0.07107855379581451
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02201734110713005,
      "epoch": 9.91,
      "learning_rate": 3.1617019410007366e-05,
      "loss": 0.0274,
      "step": 10433,
      "task_loss": 0.07534562796354294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01769842579960823,
      "epoch": 9.91,
      "learning_rate": 3.16067435713416e-05,
      "loss": 0.0172,
      "step": 10434,
      "task_loss": 0.012379692867398262
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026350311934947968,
      "epoch": 9.91,
      "learning_rate": 3.159646653252906e-05,
      "loss": 0.0298,
      "step": 10435,
      "task_loss": 0.06069865822792053
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01953766867518425,
      "epoch": 9.91,
      "learning_rate": 3.1586188295436594e-05,
      "loss": 0.0303,
      "step": 10436,
      "task_loss": 0.12754112482070923
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03570760414004326,
      "epoch": 9.91,
      "learning_rate": 3.1575908861931314e-05,
      "loss": 0.0438,
      "step": 10437,
      "task_loss": 0.11667999625205994
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018649637699127197,
      "epoch": 9.91,
      "learning_rate": 3.156562823388051e-05,
      "loss": 0.0237,
      "step": 10438,
      "task_loss": 0.06956367939710617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029894903302192688,
      "epoch": 9.91,
      "learning_rate": 3.155534641315172e-05,
      "loss": 0.0362,
      "step": 10439,
      "task_loss": 0.09323906153440475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020881684496998787,
      "epoch": 9.91,
      "learning_rate": 3.154506340161269e-05,
      "loss": 0.0296,
      "step": 10440,
      "task_loss": 0.10763251781463623
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06025876849889755,
      "epoch": 9.92,
      "learning_rate": 3.1534779201131366e-05,
      "loss": 0.0771,
      "step": 10441,
      "task_loss": 0.22887033224105835
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.059150584042072296,
      "epoch": 9.92,
      "learning_rate": 3.152449381357593e-05,
      "loss": 0.0634,
      "step": 10442,
      "task_loss": 0.1019982397556305
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08557627350091934,
      "epoch": 9.92,
      "learning_rate": 3.151420724081478e-05,
      "loss": 0.0847,
      "step": 10443,
      "task_loss": 0.07654394954442978
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05869875103235245,
      "epoch": 9.92,
      "learning_rate": 3.1503919484716495e-05,
      "loss": 0.0728,
      "step": 10444,
      "task_loss": 0.19989144802093506
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01628267765045166,
      "epoch": 9.92,
      "learning_rate": 3.149363054714992e-05,
      "loss": 0.0268,
      "step": 10445,
      "task_loss": 0.12150625884532928
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0214032344520092,
      "epoch": 9.92,
      "learning_rate": 3.148334042998408e-05,
      "loss": 0.0206,
      "step": 10446,
      "task_loss": 0.013740543276071548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0316503532230854,
      "epoch": 9.92,
      "learning_rate": 3.1473049135088225e-05,
      "loss": 0.0297,
      "step": 10447,
      "task_loss": 0.012053485959768295
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012814265675842762,
      "epoch": 9.92,
      "learning_rate": 3.146275666433183e-05,
      "loss": 0.0122,
      "step": 10448,
      "task_loss": 0.007167477160692215
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05052739381790161,
      "epoch": 9.92,
      "learning_rate": 3.145246301958455e-05,
      "loss": 0.0509,
      "step": 10449,
      "task_loss": 0.054102275520563126
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.044567592442035675,
      "epoch": 9.92,
      "learning_rate": 3.14421682027163e-05,
      "loss": 0.0422,
      "step": 10450,
      "task_loss": 0.021235253661870956
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019051125273108482,
      "epoch": 9.92,
      "learning_rate": 3.143187221559715e-05,
      "loss": 0.0181,
      "step": 10451,
      "task_loss": 0.009563138708472252
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0175587460398674,
      "epoch": 9.93,
      "learning_rate": 3.142157506009746e-05,
      "loss": 0.0162,
      "step": 10452,
      "task_loss": 0.003730185329914093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025068845599889755,
      "epoch": 9.93,
      "learning_rate": 3.141127673808772e-05,
      "loss": 0.0249,
      "step": 10453,
      "task_loss": 0.023736948147416115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012271126732230186,
      "epoch": 9.93,
      "learning_rate": 3.140097725143868e-05,
      "loss": 0.0116,
      "step": 10454,
      "task_loss": 0.005191892385482788
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0380866639316082,
      "epoch": 9.93,
      "learning_rate": 3.13906766020213e-05,
      "loss": 0.0387,
      "step": 10455,
      "task_loss": 0.04454638063907623
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0549575500190258,
      "epoch": 9.93,
      "learning_rate": 3.138037479170674e-05,
      "loss": 0.0522,
      "step": 10456,
      "task_loss": 0.02707308530807495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06924262642860413,
      "epoch": 9.93,
      "learning_rate": 3.137007182236637e-05,
      "loss": 0.0731,
      "step": 10457,
      "task_loss": 0.10814409703016281
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02348874695599079,
      "epoch": 9.93,
      "learning_rate": 3.1359767695871767e-05,
      "loss": 0.0272,
      "step": 10458,
      "task_loss": 0.06097613647580147
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04042445123195648,
      "epoch": 9.93,
      "learning_rate": 3.134946241409474e-05,
      "loss": 0.0492,
      "step": 10459,
      "task_loss": 0.127986341714859
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018123149871826172,
      "epoch": 9.93,
      "learning_rate": 3.133915597890729e-05,
      "loss": 0.02,
      "step": 10460,
      "task_loss": 0.036422938108444214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018057895824313164,
      "epoch": 9.93,
      "learning_rate": 3.132884839218162e-05,
      "loss": 0.0172,
      "step": 10461,
      "task_loss": 0.009080575779080391
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02193017117679119,
      "epoch": 9.94,
      "learning_rate": 3.131853965579016e-05,
      "loss": 0.0215,
      "step": 10462,
      "task_loss": 0.017662420868873596
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.041326068341732025,
      "epoch": 9.94,
      "learning_rate": 3.130822977160554e-05,
      "loss": 0.0484,
      "step": 10463,
      "task_loss": 0.11219672113656998
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015017848461866379,
      "epoch": 9.94,
      "learning_rate": 3.129791874150062e-05,
      "loss": 0.0244,
      "step": 10464,
      "task_loss": 0.1089889407157898
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01874970830976963,
      "epoch": 9.94,
      "learning_rate": 3.1287606567348406e-05,
      "loss": 0.0256,
      "step": 10465,
      "task_loss": 0.08684463053941727
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020482422783970833,
      "epoch": 9.94,
      "learning_rate": 3.1277293251022185e-05,
      "loss": 0.0274,
      "step": 10466,
      "task_loss": 0.08978745341300964
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02151617407798767,
      "epoch": 9.94,
      "learning_rate": 3.126697879439541e-05,
      "loss": 0.0205,
      "step": 10467,
      "task_loss": 0.011844877153635025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01606358028948307,
      "epoch": 9.94,
      "learning_rate": 3.1256663199341764e-05,
      "loss": 0.0152,
      "step": 10468,
      "task_loss": 0.00763893686234951
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030537568032741547,
      "epoch": 9.94,
      "learning_rate": 3.124634646773511e-05,
      "loss": 0.0283,
      "step": 10469,
      "task_loss": 0.007955452427268028
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02014528028666973,
      "epoch": 9.94,
      "learning_rate": 3.1236028601449534e-05,
      "loss": 0.0195,
      "step": 10470,
      "task_loss": 0.013426566496491432
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02128102444112301,
      "epoch": 9.94,
      "learning_rate": 3.1225709602359335e-05,
      "loss": 0.0285,
      "step": 10471,
      "task_loss": 0.09380625188350677
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016784438863396645,
      "epoch": 9.94,
      "learning_rate": 3.1215389472338995e-05,
      "loss": 0.0157,
      "step": 10472,
      "task_loss": 0.0061706434935331345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04040418565273285,
      "epoch": 9.95,
      "learning_rate": 3.1205068213263234e-05,
      "loss": 0.043,
      "step": 10473,
      "task_loss": 0.06682712584733963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012193303555250168,
      "epoch": 9.95,
      "learning_rate": 3.119474582700695e-05,
      "loss": 0.0168,
      "step": 10474,
      "task_loss": 0.05839819461107254
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06726071983575821,
      "epoch": 9.95,
      "learning_rate": 3.118442231544524e-05,
      "loss": 0.071,
      "step": 10475,
      "task_loss": 0.10484815388917923
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026226460933685303,
      "epoch": 9.95,
      "learning_rate": 3.117409768045344e-05,
      "loss": 0.0296,
      "step": 10476,
      "task_loss": 0.06030692532658577
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011876748874783516,
      "epoch": 9.95,
      "learning_rate": 3.116377192390706e-05,
      "loss": 0.0113,
      "step": 10477,
      "task_loss": 0.006212221458554268
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08747504651546478,
      "epoch": 9.95,
      "learning_rate": 3.115344504768183e-05,
      "loss": 0.0951,
      "step": 10478,
      "task_loss": 0.16354155540466309
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07612023502588272,
      "epoch": 9.95,
      "learning_rate": 3.1143117053653665e-05,
      "loss": 0.0816,
      "step": 10479,
      "task_loss": 0.1306367814540863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02969006821513176,
      "epoch": 9.95,
      "learning_rate": 3.113278794369869e-05,
      "loss": 0.0357,
      "step": 10480,
      "task_loss": 0.08966681361198425
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08898676186800003,
      "epoch": 9.95,
      "learning_rate": 3.112245771969327e-05,
      "loss": 0.0848,
      "step": 10481,
      "task_loss": 0.04718249663710594
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11488527804613113,
      "epoch": 9.95,
      "learning_rate": 3.1112126383513914e-05,
      "loss": 0.1121,
      "step": 10482,
      "task_loss": 0.08729420602321625
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.048652954399585724,
      "epoch": 9.96,
      "learning_rate": 3.110179393703737e-05,
      "loss": 0.0742,
      "step": 10483,
      "task_loss": 0.30367979407310486
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019496137276291847,
      "epoch": 9.96,
      "learning_rate": 3.109146038214055e-05,
      "loss": 0.0269,
      "step": 10484,
      "task_loss": 0.09309974312782288
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01703246682882309,
      "epoch": 9.96,
      "learning_rate": 3.108112572070063e-05,
      "loss": 0.0197,
      "step": 10485,
      "task_loss": 0.043584369122982025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05471110716462135,
      "epoch": 9.96,
      "learning_rate": 3.1070789954594934e-05,
      "loss": 0.0517,
      "step": 10486,
      "task_loss": 0.024789560586214066
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02847129851579666,
      "epoch": 9.96,
      "learning_rate": 3.1060453085701e-05,
      "loss": 0.0314,
      "step": 10487,
      "task_loss": 0.05726098641753197
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018717454746365547,
      "epoch": 9.96,
      "learning_rate": 3.105011511589658e-05,
      "loss": 0.0232,
      "step": 10488,
      "task_loss": 0.06356090307235718
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030354956164956093,
      "epoch": 9.96,
      "learning_rate": 3.103977604705961e-05,
      "loss": 0.0287,
      "step": 10489,
      "task_loss": 0.013773368671536446
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05581222474575043,
      "epoch": 9.96,
      "learning_rate": 3.102943588106824e-05,
      "loss": 0.053,
      "step": 10490,
      "task_loss": 0.027538809925317764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021054701879620552,
      "epoch": 9.96,
      "learning_rate": 3.10190946198008e-05,
      "loss": 0.0211,
      "step": 10491,
      "task_loss": 0.021210404112935066
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.041227295994758606,
      "epoch": 9.96,
      "learning_rate": 3.100875226513583e-05,
      "loss": 0.0386,
      "step": 10492,
      "task_loss": 0.015084227547049522
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03447715565562248,
      "epoch": 9.96,
      "learning_rate": 3.099840881895208e-05,
      "loss": 0.0406,
      "step": 10493,
      "task_loss": 0.09537314623594284
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05250038951635361,
      "epoch": 9.97,
      "learning_rate": 3.098806428312847e-05,
      "loss": 0.0565,
      "step": 10494,
      "task_loss": 0.09219056367874146
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0351855605840683,
      "epoch": 9.97,
      "learning_rate": 3.097771865954415e-05,
      "loss": 0.036,
      "step": 10495,
      "task_loss": 0.04331498593091965
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0283675417304039,
      "epoch": 9.97,
      "learning_rate": 3.096737195007845e-05,
      "loss": 0.0341,
      "step": 10496,
      "task_loss": 0.08616603910923004
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029870422556996346,
      "epoch": 9.97,
      "learning_rate": 3.0957024156610884e-05,
      "loss": 0.038,
      "step": 10497,
      "task_loss": 0.11162212491035461
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012384091503918171,
      "epoch": 9.97,
      "learning_rate": 3.09466752810212e-05,
      "loss": 0.0158,
      "step": 10498,
      "task_loss": 0.04627315700054169
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013409021310508251,
      "epoch": 9.97,
      "learning_rate": 3.093632532518931e-05,
      "loss": 0.0155,
      "step": 10499,
      "task_loss": 0.03382343798875809
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010947839356958866,
      "epoch": 9.97,
      "learning_rate": 3.092597429099534e-05,
      "loss": 0.0137,
      "step": 10500,
      "task_loss": 0.03889209032058716
    },
    {
      "epoch": 9.97,
      "eval_accuracy": 0.8956422018348624,
      "eval_loss": 0.4390338361263275,
      "eval_runtime": 18.2957,
      "eval_samples_per_second": 47.661,
      "eval_steps_per_second": 5.958,
      "step": 10500
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02746719866991043,
      "epoch": 9.97,
      "learning_rate": 3.0915622180319585e-05,
      "loss": 0.0316,
      "step": 10501,
      "task_loss": 0.06869719922542572
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.039203185588121414,
      "epoch": 9.97,
      "learning_rate": 3.090526899504259e-05,
      "loss": 0.0372,
      "step": 10502,
      "task_loss": 0.019136063754558563
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025588029995560646,
      "epoch": 9.97,
      "learning_rate": 3.0894914737045034e-05,
      "loss": 0.0305,
      "step": 10503,
      "task_loss": 0.07455385476350784
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030900314450263977,
      "epoch": 9.98,
      "learning_rate": 3.088455940820782e-05,
      "loss": 0.0378,
      "step": 10504,
      "task_loss": 0.09984858334064484
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.059486109763383865,
      "epoch": 9.98,
      "learning_rate": 3.087420301041206e-05,
      "loss": 0.0644,
      "step": 10505,
      "task_loss": 0.10867089033126831
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012473450042307377,
      "epoch": 9.98,
      "learning_rate": 3.086384554553902e-05,
      "loss": 0.0157,
      "step": 10506,
      "task_loss": 0.044721417129039764
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.037564076483249664,
      "epoch": 9.98,
      "learning_rate": 3.0853487015470206e-05,
      "loss": 0.0342,
      "step": 10507,
      "task_loss": 0.00427047535777092
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.21402209997177124,
      "epoch": 9.98,
      "learning_rate": 3.084312742208728e-05,
      "loss": 0.2225,
      "step": 10508,
      "task_loss": 0.2992434501647949
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05252481997013092,
      "epoch": 9.98,
      "learning_rate": 3.083276676727212e-05,
      "loss": 0.0562,
      "step": 10509,
      "task_loss": 0.08937288820743561
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02034352719783783,
      "epoch": 9.98,
      "learning_rate": 3.082240505290678e-05,
      "loss": 0.0345,
      "step": 10510,
      "task_loss": 0.16238275170326233
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02308393456041813,
      "epoch": 9.98,
      "learning_rate": 3.081204228087353e-05,
      "loss": 0.0238,
      "step": 10511,
      "task_loss": 0.02994043007493019
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022067924961447716,
      "epoch": 9.98,
      "learning_rate": 3.08016784530548e-05,
      "loss": 0.0206,
      "step": 10512,
      "task_loss": 0.007659193128347397
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017336195334792137,
      "epoch": 9.98,
      "learning_rate": 3.0791313571333244e-05,
      "loss": 0.0188,
      "step": 10513,
      "task_loss": 0.031775638461112976
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04067476838827133,
      "epoch": 9.98,
      "learning_rate": 3.078094763759168e-05,
      "loss": 0.0461,
      "step": 10514,
      "task_loss": 0.09492932260036469
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01405811682343483,
      "epoch": 9.99,
      "learning_rate": 3.0770580653713146e-05,
      "loss": 0.0137,
      "step": 10515,
      "task_loss": 0.010203549638390541
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023571021854877472,
      "epoch": 9.99,
      "learning_rate": 3.076021262158084e-05,
      "loss": 0.038,
      "step": 10516,
      "task_loss": 0.1676616072654724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019903654232621193,
      "epoch": 9.99,
      "learning_rate": 3.074984354307817e-05,
      "loss": 0.0192,
      "step": 10517,
      "task_loss": 0.012664202600717545
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0405997633934021,
      "epoch": 9.99,
      "learning_rate": 3.073947342008873e-05,
      "loss": 0.0409,
      "step": 10518,
      "task_loss": 0.04333402216434479
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022368015721440315,
      "epoch": 9.99,
      "learning_rate": 3.07291022544963e-05,
      "loss": 0.0232,
      "step": 10519,
      "task_loss": 0.030709004029631615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013003375381231308,
      "epoch": 9.99,
      "learning_rate": 3.0718730048184855e-05,
      "loss": 0.0123,
      "step": 10520,
      "task_loss": 0.0056927260011434555
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021634429693222046,
      "epoch": 9.99,
      "learning_rate": 3.0708356803038556e-05,
      "loss": 0.0208,
      "step": 10521,
      "task_loss": 0.01319202035665512
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026351455599069595,
      "epoch": 9.99,
      "learning_rate": 3.069798252094175e-05,
      "loss": 0.0246,
      "step": 10522,
      "task_loss": 0.008785083889961243
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08953380584716797,
      "epoch": 9.99,
      "learning_rate": 3.068760720377897e-05,
      "loss": 0.0882,
      "step": 10523,
      "task_loss": 0.07605984061956406
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01611064001917839,
      "epoch": 9.99,
      "learning_rate": 3.067723085343496e-05,
      "loss": 0.0154,
      "step": 10524,
      "task_loss": 0.008628584444522858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03214561194181442,
      "epoch": 10.0,
      "learning_rate": 3.066685347179462e-05,
      "loss": 0.0427,
      "step": 10525,
      "task_loss": 0.13809731602668762
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017522085458040237,
      "epoch": 10.0,
      "learning_rate": 3.065647506074306e-05,
      "loss": 0.0165,
      "step": 10526,
      "task_loss": 0.007697628811001778
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03798111528158188,
      "epoch": 10.0,
      "learning_rate": 3.064609562216555e-05,
      "loss": 0.0371,
      "step": 10527,
      "task_loss": 0.028712373226881027
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08762382715940475,
      "epoch": 10.0,
      "learning_rate": 3.063571515794759e-05,
      "loss": 0.09,
      "step": 10528,
      "task_loss": 0.11169522255659103
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.060012731701135635,
      "epoch": 10.0,
      "learning_rate": 3.062533366997483e-05,
      "loss": 0.0631,
      "step": 10529,
      "task_loss": 0.09081701934337616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06982729583978653,
      "epoch": 10.0,
      "learning_rate": 3.061495116013311e-05,
      "loss": 0.0848,
      "step": 10530,
      "task_loss": 0.21950267255306244
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015027199871838093,
      "epoch": 10.0,
      "learning_rate": 3.060456763030847e-05,
      "loss": 0.0151,
      "step": 10531,
      "task_loss": 0.015308814123272896
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010798173025250435,
      "epoch": 10.0,
      "learning_rate": 3.059418308238713e-05,
      "loss": 0.011,
      "step": 10532,
      "task_loss": 0.012500785291194916
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03574404492974281,
      "epoch": 10.0,
      "learning_rate": 3.0583797518255505e-05,
      "loss": 0.0326,
      "step": 10533,
      "task_loss": 0.0042223744094371796
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021271761506795883,
      "epoch": 10.0,
      "learning_rate": 3.057341093980015e-05,
      "loss": 0.0316,
      "step": 10534,
      "task_loss": 0.12449042499065399
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11009443551301956,
      "epoch": 10.0,
      "learning_rate": 3.056302334890786e-05,
      "loss": 0.1068,
      "step": 10535,
      "task_loss": 0.07683957368135452
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018123319372534752,
      "epoch": 10.01,
      "learning_rate": 3.055263474746559e-05,
      "loss": 0.0185,
      "step": 10536,
      "task_loss": 0.021713176742196083
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016921810805797577,
      "epoch": 10.01,
      "learning_rate": 3.054224513736048e-05,
      "loss": 0.0374,
      "step": 10537,
      "task_loss": 0.2221601903438568
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017352506518363953,
      "epoch": 10.01,
      "learning_rate": 3.0531854520479844e-05,
      "loss": 0.0216,
      "step": 10538,
      "task_loss": 0.05965179204940796
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04468035697937012,
      "epoch": 10.01,
      "learning_rate": 3.05214628987112e-05,
      "loss": 0.0476,
      "step": 10539,
      "task_loss": 0.07353688031435013
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015710055828094482,
      "epoch": 10.01,
      "learning_rate": 3.0511070273942217e-05,
      "loss": 0.0148,
      "step": 10540,
      "task_loss": 0.006397552788257599
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014855926856398582,
      "epoch": 10.01,
      "learning_rate": 3.0500676648060776e-05,
      "loss": 0.0181,
      "step": 10541,
      "task_loss": 0.046922795474529266
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019756915047764778,
      "epoch": 10.01,
      "learning_rate": 3.049028202295494e-05,
      "loss": 0.0209,
      "step": 10542,
      "task_loss": 0.03078971616923809
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014726649969816208,
      "epoch": 10.01,
      "learning_rate": 3.0479886400512937e-05,
      "loss": 0.0273,
      "step": 10543,
      "task_loss": 0.14058496057987213
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030262373387813568,
      "epoch": 10.01,
      "learning_rate": 3.0469489782623163e-05,
      "loss": 0.029,
      "step": 10544,
      "task_loss": 0.01749301515519619
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024717876687645912,
      "epoch": 10.01,
      "learning_rate": 3.045909217117424e-05,
      "loss": 0.0291,
      "step": 10545,
      "task_loss": 0.06830822676420212
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010806870646774769,
      "epoch": 10.02,
      "learning_rate": 3.0448693568054924e-05,
      "loss": 0.0103,
      "step": 10546,
      "task_loss": 0.005712360143661499
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022307565435767174,
      "epoch": 10.02,
      "learning_rate": 3.0438293975154186e-05,
      "loss": 0.0247,
      "step": 10547,
      "task_loss": 0.04641730338335037
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.031023293733596802,
      "epoch": 10.02,
      "learning_rate": 3.042789339436116e-05,
      "loss": 0.0311,
      "step": 10548,
      "task_loss": 0.031546495854854584
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02841011807322502,
      "epoch": 10.02,
      "learning_rate": 3.041749182756515e-05,
      "loss": 0.0366,
      "step": 10549,
      "task_loss": 0.11078496277332306
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03763877600431442,
      "epoch": 10.02,
      "learning_rate": 3.0407089276655664e-05,
      "loss": 0.0385,
      "step": 10550,
      "task_loss": 0.04576456546783447
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01621590554714203,
      "epoch": 10.02,
      "learning_rate": 3.039668574352237e-05,
      "loss": 0.027,
      "step": 10551,
      "task_loss": 0.12407008558511734
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018488582223653793,
      "epoch": 10.02,
      "learning_rate": 3.0386281230055113e-05,
      "loss": 0.0176,
      "step": 10552,
      "task_loss": 0.009799566119909286
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012251723557710648,
      "epoch": 10.02,
      "learning_rate": 3.0375875738143938e-05,
      "loss": 0.0197,
      "step": 10553,
      "task_loss": 0.08636859059333801
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019306058064103127,
      "epoch": 10.02,
      "learning_rate": 3.0365469269679042e-05,
      "loss": 0.0194,
      "step": 10554,
      "task_loss": 0.020166553556919098
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022004347294569016,
      "epoch": 10.02,
      "learning_rate": 3.0355061826550813e-05,
      "loss": 0.0204,
      "step": 10555,
      "task_loss": 0.005572935566306114
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010447701439261436,
      "epoch": 10.02,
      "learning_rate": 3.0344653410649815e-05,
      "loss": 0.0171,
      "step": 10556,
      "task_loss": 0.07670579850673676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023317281156778336,
      "epoch": 10.03,
      "learning_rate": 3.033424402386678e-05,
      "loss": 0.0288,
      "step": 10557,
      "task_loss": 0.07775168120861053
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01650930941104889,
      "epoch": 10.03,
      "learning_rate": 3.032383366809263e-05,
      "loss": 0.0173,
      "step": 10558,
      "task_loss": 0.02475823275744915
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016953203827142715,
      "epoch": 10.03,
      "learning_rate": 3.031342234521845e-05,
      "loss": 0.016,
      "step": 10559,
      "task_loss": 0.007815249264240265
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.045376237481832504,
      "epoch": 10.03,
      "learning_rate": 3.030301005713552e-05,
      "loss": 0.0476,
      "step": 10560,
      "task_loss": 0.06719513982534409
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010588767006993294,
      "epoch": 10.03,
      "learning_rate": 3.0292596805735274e-05,
      "loss": 0.0157,
      "step": 10561,
      "task_loss": 0.06145908683538437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013043254613876343,
      "epoch": 10.03,
      "learning_rate": 3.028218259290932e-05,
      "loss": 0.0121,
      "step": 10562,
      "task_loss": 0.003523891791701317
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030626816675066948,
      "epoch": 10.03,
      "learning_rate": 3.0271767420549463e-05,
      "loss": 0.0305,
      "step": 10563,
      "task_loss": 0.02899116836488247
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015171288512647152,
      "epoch": 10.03,
      "learning_rate": 3.0261351290547667e-05,
      "loss": 0.019,
      "step": 10564,
      "task_loss": 0.053528182208538055
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011217884719371796,
      "epoch": 10.03,
      "learning_rate": 3.025093420479607e-05,
      "loss": 0.0113,
      "step": 10565,
      "task_loss": 0.012106716632843018
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.055433209985494614,
      "epoch": 10.03,
      "learning_rate": 3.0240516165186976e-05,
      "loss": 0.055,
      "step": 10566,
      "task_loss": 0.05153985321521759
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020901085808873177,
      "epoch": 10.04,
      "learning_rate": 3.0230097173612896e-05,
      "loss": 0.0282,
      "step": 10567,
      "task_loss": 0.09422680735588074
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.00867534801363945,
      "epoch": 10.04,
      "learning_rate": 3.021967723196647e-05,
      "loss": 0.0227,
      "step": 10568,
      "task_loss": 0.1493276208639145
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02514798566699028,
      "epoch": 10.04,
      "learning_rate": 3.020925634214054e-05,
      "loss": 0.0274,
      "step": 10569,
      "task_loss": 0.04775901138782501
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010010787285864353,
      "epoch": 10.04,
      "learning_rate": 3.01988345060281e-05,
      "loss": 0.0094,
      "step": 10570,
      "task_loss": 0.0035102032124996185
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03369935229420662,
      "epoch": 10.04,
      "learning_rate": 3.018841172552234e-05,
      "loss": 0.033,
      "step": 10571,
      "task_loss": 0.026725368574261665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06951646506786346,
      "epoch": 10.04,
      "learning_rate": 3.01779880025166e-05,
      "loss": 0.073,
      "step": 10572,
      "task_loss": 0.10399917513132095
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016049357131123543,
      "epoch": 10.04,
      "learning_rate": 3.0167563338904402e-05,
      "loss": 0.0153,
      "step": 10573,
      "task_loss": 0.008852284401655197
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024919727817177773,
      "epoch": 10.04,
      "learning_rate": 3.0157137736579445e-05,
      "loss": 0.024,
      "step": 10574,
      "task_loss": 0.01523817703127861
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01857706718146801,
      "epoch": 10.04,
      "learning_rate": 3.014671119743556e-05,
      "loss": 0.0173,
      "step": 10575,
      "task_loss": 0.0053652990609407425
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016847968101501465,
      "epoch": 10.04,
      "learning_rate": 3.013628372336682e-05,
      "loss": 0.0155,
      "step": 10576,
      "task_loss": 0.0033319760113954544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018795382231473923,
      "epoch": 10.04,
      "learning_rate": 3.0125855316267394e-05,
      "loss": 0.0242,
      "step": 10577,
      "task_loss": 0.07253532111644745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06431741267442703,
      "epoch": 10.05,
      "learning_rate": 3.0115425978031663e-05,
      "loss": 0.0702,
      "step": 10578,
      "task_loss": 0.12317221611738205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.047512613236904144,
      "epoch": 10.05,
      "learning_rate": 3.0104995710554174e-05,
      "loss": 0.0457,
      "step": 10579,
      "task_loss": 0.02974691055715084
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07968870550394058,
      "epoch": 10.05,
      "learning_rate": 3.0094564515729623e-05,
      "loss": 0.0936,
      "step": 10580,
      "task_loss": 0.21833162009716034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008304203860461712,
      "epoch": 10.05,
      "learning_rate": 3.0084132395452896e-05,
      "loss": 0.0079,
      "step": 10581,
      "task_loss": 0.004243377596139908
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016128698363900185,
      "epoch": 10.05,
      "learning_rate": 3.0073699351619033e-05,
      "loss": 0.0175,
      "step": 10582,
      "task_loss": 0.03016304410994053
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01745760627090931,
      "epoch": 10.05,
      "learning_rate": 3.0063265386123247e-05,
      "loss": 0.0251,
      "step": 10583,
      "task_loss": 0.09426887333393097
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023326557129621506,
      "epoch": 10.05,
      "learning_rate": 3.0052830500860912e-05,
      "loss": 0.0215,
      "step": 10584,
      "task_loss": 0.004915602505207062
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04375822842121124,
      "epoch": 10.05,
      "learning_rate": 3.0042394697727587e-05,
      "loss": 0.0401,
      "step": 10585,
      "task_loss": 0.007045827805995941
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012805687263607979,
      "epoch": 10.05,
      "learning_rate": 3.0031957978618986e-05,
      "loss": 0.012,
      "step": 10586,
      "task_loss": 0.005244376137852669
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01600104570388794,
      "epoch": 10.05,
      "learning_rate": 3.002152034543098e-05,
      "loss": 0.015,
      "step": 10587,
      "task_loss": 0.005925571545958519
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008856108412146568,
      "epoch": 10.06,
      "learning_rate": 3.0011081800059616e-05,
      "loss": 0.0084,
      "step": 10588,
      "task_loss": 0.003961298614740372
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02766408771276474,
      "epoch": 10.06,
      "learning_rate": 3.0000642344401113e-05,
      "loss": 0.0406,
      "step": 10589,
      "task_loss": 0.15749169886112213
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014702584594488144,
      "epoch": 10.06,
      "learning_rate": 2.999020198035184e-05,
      "loss": 0.0137,
      "step": 10590,
      "task_loss": 0.004804585129022598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013725142925977707,
      "epoch": 10.06,
      "learning_rate": 2.997976070980836e-05,
      "loss": 0.0143,
      "step": 10591,
      "task_loss": 0.019009843468666077
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027275511994957924,
      "epoch": 10.06,
      "learning_rate": 2.996931853466734e-05,
      "loss": 0.0451,
      "step": 10592,
      "task_loss": 0.20579108595848083
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009938783943653107,
      "epoch": 10.06,
      "learning_rate": 2.9958875456825692e-05,
      "loss": 0.0093,
      "step": 10593,
      "task_loss": 0.003395296633243561
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.00878879614174366,
      "epoch": 10.06,
      "learning_rate": 2.9948431478180434e-05,
      "loss": 0.01,
      "step": 10594,
      "task_loss": 0.021191343665122986
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017414798960089684,
      "epoch": 10.06,
      "learning_rate": 2.9937986600628758e-05,
      "loss": 0.0165,
      "step": 10595,
      "task_loss": 0.008245648816227913
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008904863148927689,
      "epoch": 10.06,
      "learning_rate": 2.992754082606804e-05,
      "loss": 0.0085,
      "step": 10596,
      "task_loss": 0.004817705601453781
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05227641388773918,
      "epoch": 10.06,
      "learning_rate": 2.9917094156395796e-05,
      "loss": 0.0602,
      "step": 10597,
      "task_loss": 0.1313791424036026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016239894554018974,
      "epoch": 10.06,
      "learning_rate": 2.990664659350973e-05,
      "loss": 0.0153,
      "step": 10598,
      "task_loss": 0.006773691624403
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028258735314011574,
      "epoch": 10.07,
      "learning_rate": 2.9896198139307668e-05,
      "loss": 0.0283,
      "step": 10599,
      "task_loss": 0.028914695605635643
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027807846665382385,
      "epoch": 10.07,
      "learning_rate": 2.9885748795687642e-05,
      "loss": 0.0382,
      "step": 10600,
      "task_loss": 0.1314897984266281
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07431833446025848,
      "epoch": 10.07,
      "learning_rate": 2.9875298564547805e-05,
      "loss": 0.0764,
      "step": 10601,
      "task_loss": 0.09555675089359283
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03642863780260086,
      "epoch": 10.07,
      "learning_rate": 2.9864847447786503e-05,
      "loss": 0.0349,
      "step": 10602,
      "task_loss": 0.021108128130435944
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030721113085746765,
      "epoch": 10.07,
      "learning_rate": 2.9854395447302246e-05,
      "loss": 0.0358,
      "step": 10603,
      "task_loss": 0.08126336336135864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03847306966781616,
      "epoch": 10.07,
      "learning_rate": 2.9843942564993672e-05,
      "loss": 0.0412,
      "step": 10604,
      "task_loss": 0.06578797101974487
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.007311783730983734,
      "epoch": 10.07,
      "learning_rate": 2.98334888027596e-05,
      "loss": 0.007,
      "step": 10605,
      "task_loss": 0.004128355532884598
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017765909433364868,
      "epoch": 10.07,
      "learning_rate": 2.9823034162499007e-05,
      "loss": 0.0178,
      "step": 10606,
      "task_loss": 0.017818637192249298
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008372337557375431,
      "epoch": 10.07,
      "learning_rate": 2.981257864611104e-05,
      "loss": 0.0081,
      "step": 10607,
      "task_loss": 0.0059828683733940125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0075715347193181515,
      "epoch": 10.07,
      "learning_rate": 2.980212225549498e-05,
      "loss": 0.0072,
      "step": 10608,
      "task_loss": 0.0034196972846984863
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015017621219158173,
      "epoch": 10.08,
      "learning_rate": 2.9791664992550273e-05,
      "loss": 0.016,
      "step": 10609,
      "task_loss": 0.024808872491121292
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03149663656949997,
      "epoch": 10.08,
      "learning_rate": 2.978120685917656e-05,
      "loss": 0.0478,
      "step": 10610,
      "task_loss": 0.19493995606899261
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018071962520480156,
      "epoch": 10.08,
      "learning_rate": 2.9770747857273584e-05,
      "loss": 0.0249,
      "step": 10611,
      "task_loss": 0.08605194836854935
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020836105570197105,
      "epoch": 10.08,
      "learning_rate": 2.9760287988741293e-05,
      "loss": 0.0341,
      "step": 10612,
      "task_loss": 0.15324726700782776
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.11783700436353683,
      "epoch": 10.08,
      "learning_rate": 2.9749827255479755e-05,
      "loss": 0.1251,
      "step": 10613,
      "task_loss": 0.19090542197227478
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05260676145553589,
      "epoch": 10.08,
      "learning_rate": 2.9739365659389223e-05,
      "loss": 0.0506,
      "step": 10614,
      "task_loss": 0.03276871144771576
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020378753542900085,
      "epoch": 10.08,
      "learning_rate": 2.972890320237009e-05,
      "loss": 0.0242,
      "step": 10615,
      "task_loss": 0.05864740163087845
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027730267494916916,
      "epoch": 10.08,
      "learning_rate": 2.971843988632292e-05,
      "loss": 0.0394,
      "step": 10616,
      "task_loss": 0.14460690319538116
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009068669751286507,
      "epoch": 10.08,
      "learning_rate": 2.970797571314842e-05,
      "loss": 0.0085,
      "step": 10617,
      "task_loss": 0.0035065151751041412
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014962357468903065,
      "epoch": 10.08,
      "learning_rate": 2.9697510684747454e-05,
      "loss": 0.0271,
      "step": 10618,
      "task_loss": 0.13601535558700562
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0636109784245491,
      "epoch": 10.08,
      "learning_rate": 2.9687044803021057e-05,
      "loss": 0.0619,
      "step": 10619,
      "task_loss": 0.046097345650196075
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01265124510973692,
      "epoch": 10.09,
      "learning_rate": 2.9676578069870392e-05,
      "loss": 0.0118,
      "step": 10620,
      "task_loss": 0.004545770585536957
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02781161107122898,
      "epoch": 10.09,
      "learning_rate": 2.9666110487196798e-05,
      "loss": 0.0261,
      "step": 10621,
      "task_loss": 0.011152038350701332
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03320011496543884,
      "epoch": 10.09,
      "learning_rate": 2.9655642056901762e-05,
      "loss": 0.0364,
      "step": 10622,
      "task_loss": 0.06528480350971222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08747844398021698,
      "epoch": 10.09,
      "learning_rate": 2.9645172780886927e-05,
      "loss": 0.092,
      "step": 10623,
      "task_loss": 0.13289803266525269
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024324875324964523,
      "epoch": 10.09,
      "learning_rate": 2.9634702661054085e-05,
      "loss": 0.023,
      "step": 10624,
      "task_loss": 0.011415783315896988
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009565697982907295,
      "epoch": 10.09,
      "learning_rate": 2.962423169930518e-05,
      "loss": 0.0089,
      "step": 10625,
      "task_loss": 0.003159165382385254
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013625957071781158,
      "epoch": 10.09,
      "learning_rate": 2.961375989754232e-05,
      "loss": 0.0128,
      "step": 10626,
      "task_loss": 0.0050227586179971695
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05235249549150467,
      "epoch": 10.09,
      "learning_rate": 2.9603287257667754e-05,
      "loss": 0.0643,
      "step": 10627,
      "task_loss": 0.17228303849697113
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021474594250321388,
      "epoch": 10.09,
      "learning_rate": 2.9592813781583885e-05,
      "loss": 0.02,
      "step": 10628,
      "task_loss": 0.007135756313800812
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027554640546441078,
      "epoch": 10.09,
      "learning_rate": 2.958233947119328e-05,
      "loss": 0.0254,
      "step": 10629,
      "task_loss": 0.005915815010666847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06590898334980011,
      "epoch": 10.09,
      "learning_rate": 2.9571864328398636e-05,
      "loss": 0.0754,
      "step": 10630,
      "task_loss": 0.1609448492527008
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04161786288022995,
      "epoch": 10.1,
      "learning_rate": 2.956138835510282e-05,
      "loss": 0.0464,
      "step": 10631,
      "task_loss": 0.08935808390378952
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01830127462744713,
      "epoch": 10.1,
      "learning_rate": 2.9550911553208838e-05,
      "loss": 0.02,
      "step": 10632,
      "task_loss": 0.034862592816352844
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0464775525033474,
      "epoch": 10.1,
      "learning_rate": 2.954043392461986e-05,
      "loss": 0.0431,
      "step": 10633,
      "task_loss": 0.01255410723388195
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021407488733530045,
      "epoch": 10.1,
      "learning_rate": 2.952995547123919e-05,
      "loss": 0.0287,
      "step": 10634,
      "task_loss": 0.09397697448730469
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013649847358465195,
      "epoch": 10.1,
      "learning_rate": 2.9519476194970286e-05,
      "loss": 0.0128,
      "step": 10635,
      "task_loss": 0.005092758685350418
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03821259364485741,
      "epoch": 10.1,
      "learning_rate": 2.9508996097716777e-05,
      "loss": 0.0355,
      "step": 10636,
      "task_loss": 0.011517934501171112
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0182490237057209,
      "epoch": 10.1,
      "learning_rate": 2.949851518138241e-05,
      "loss": 0.0306,
      "step": 10637,
      "task_loss": 0.14180639386177063
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.031840305775403976,
      "epoch": 10.1,
      "learning_rate": 2.948803344787109e-05,
      "loss": 0.0356,
      "step": 10638,
      "task_loss": 0.06947970390319824
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01208855863660574,
      "epoch": 10.1,
      "learning_rate": 2.947755089908688e-05,
      "loss": 0.0287,
      "step": 10639,
      "task_loss": 0.17781664431095123
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01693141832947731,
      "epoch": 10.1,
      "learning_rate": 2.946706753693398e-05,
      "loss": 0.0278,
      "step": 10640,
      "task_loss": 0.1261083036661148
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014154396951198578,
      "epoch": 10.11,
      "learning_rate": 2.945658336331676e-05,
      "loss": 0.0231,
      "step": 10641,
      "task_loss": 0.10346969962120056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0599503219127655,
      "epoch": 10.11,
      "learning_rate": 2.9446098380139703e-05,
      "loss": 0.0846,
      "step": 10642,
      "task_loss": 0.30606013536453247
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02652948535978794,
      "epoch": 10.11,
      "learning_rate": 2.9435612589307458e-05,
      "loss": 0.0424,
      "step": 10643,
      "task_loss": 0.18499121069908142
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028400473296642303,
      "epoch": 10.11,
      "learning_rate": 2.942512599272483e-05,
      "loss": 0.0444,
      "step": 10644,
      "task_loss": 0.18858596682548523
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02752113714814186,
      "epoch": 10.11,
      "learning_rate": 2.9414638592296752e-05,
      "loss": 0.031,
      "step": 10645,
      "task_loss": 0.06193367764353752
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02233045920729637,
      "epoch": 10.11,
      "learning_rate": 2.9404150389928316e-05,
      "loss": 0.039,
      "step": 10646,
      "task_loss": 0.18857558071613312
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012071599252521992,
      "epoch": 10.11,
      "learning_rate": 2.9393661387524745e-05,
      "loss": 0.0204,
      "step": 10647,
      "task_loss": 0.0955902636051178
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009622898884117603,
      "epoch": 10.11,
      "learning_rate": 2.9383171586991424e-05,
      "loss": 0.0171,
      "step": 10648,
      "task_loss": 0.08440146595239639
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011028185486793518,
      "epoch": 10.11,
      "learning_rate": 2.9372680990233875e-05,
      "loss": 0.0179,
      "step": 10649,
      "task_loss": 0.07927846908569336
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0299701951444149,
      "epoch": 10.11,
      "learning_rate": 2.9362189599157776e-05,
      "loss": 0.038,
      "step": 10650,
      "task_loss": 0.10991066694259644
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01364204753190279,
      "epoch": 10.11,
      "learning_rate": 2.9351697415668917e-05,
      "loss": 0.0286,
      "step": 10651,
      "task_loss": 0.16349458694458008
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014143523760139942,
      "epoch": 10.12,
      "learning_rate": 2.9341204441673266e-05,
      "loss": 0.0235,
      "step": 10652,
      "task_loss": 0.10781119763851166
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0211515910923481,
      "epoch": 10.12,
      "learning_rate": 2.9330710679076916e-05,
      "loss": 0.0267,
      "step": 10653,
      "task_loss": 0.07670915871858597
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011677171103656292,
      "epoch": 10.12,
      "learning_rate": 2.9320216129786116e-05,
      "loss": 0.0113,
      "step": 10654,
      "task_loss": 0.007633142173290253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03487570211291313,
      "epoch": 10.12,
      "learning_rate": 2.9309720795707257e-05,
      "loss": 0.0425,
      "step": 10655,
      "task_loss": 0.11129167675971985
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014407580718398094,
      "epoch": 10.12,
      "learning_rate": 2.9299224678746855e-05,
      "loss": 0.0138,
      "step": 10656,
      "task_loss": 0.008088191971182823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014523474499583244,
      "epoch": 10.12,
      "learning_rate": 2.928872778081158e-05,
      "loss": 0.0136,
      "step": 10657,
      "task_loss": 0.005704553797841072
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05787359178066254,
      "epoch": 10.12,
      "learning_rate": 2.9278230103808257e-05,
      "loss": 0.0582,
      "step": 10658,
      "task_loss": 0.061048611998558044
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020860906690359116,
      "epoch": 10.12,
      "learning_rate": 2.9267731649643827e-05,
      "loss": 0.0373,
      "step": 10659,
      "task_loss": 0.18554821610450745
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015227505937218666,
      "epoch": 10.12,
      "learning_rate": 2.9257232420225394e-05,
      "loss": 0.0218,
      "step": 10660,
      "task_loss": 0.08097986876964569
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013915725983679295,
      "epoch": 10.12,
      "learning_rate": 2.9246732417460178e-05,
      "loss": 0.0131,
      "step": 10661,
      "task_loss": 0.005335215479135513
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015429834835231304,
      "epoch": 10.13,
      "learning_rate": 2.9236231643255578e-05,
      "loss": 0.0209,
      "step": 10662,
      "task_loss": 0.07019021362066269
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016493186354637146,
      "epoch": 10.13,
      "learning_rate": 2.922573009951909e-05,
      "loss": 0.0259,
      "step": 10663,
      "task_loss": 0.11039119958877563
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016565848141908646,
      "epoch": 10.13,
      "learning_rate": 2.9215227788158382e-05,
      "loss": 0.0183,
      "step": 10664,
      "task_loss": 0.03420030698180199
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021225640550255775,
      "epoch": 10.13,
      "learning_rate": 2.920472471108125e-05,
      "loss": 0.0269,
      "step": 10665,
      "task_loss": 0.07766622304916382
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06270953267812729,
      "epoch": 10.13,
      "learning_rate": 2.919422087019561e-05,
      "loss": 0.0811,
      "step": 10666,
      "task_loss": 0.24679216742515564
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.00949104130268097,
      "epoch": 10.13,
      "learning_rate": 2.9183716267409562e-05,
      "loss": 0.0097,
      "step": 10667,
      "task_loss": 0.011834444478154182
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016700657084584236,
      "epoch": 10.13,
      "learning_rate": 2.9173210904631297e-05,
      "loss": 0.0192,
      "step": 10668,
      "task_loss": 0.04135492444038391
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019157694652676582,
      "epoch": 10.13,
      "learning_rate": 2.916270478376918e-05,
      "loss": 0.018,
      "step": 10669,
      "task_loss": 0.00787227414548397
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015583924949169159,
      "epoch": 10.13,
      "learning_rate": 2.9152197906731687e-05,
      "loss": 0.0191,
      "step": 10670,
      "task_loss": 0.05046524479985237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0134699996560812,
      "epoch": 10.13,
      "learning_rate": 2.9141690275427445e-05,
      "loss": 0.0137,
      "step": 10671,
      "task_loss": 0.015451043844223022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023017989471554756,
      "epoch": 10.13,
      "learning_rate": 2.9131181891765226e-05,
      "loss": 0.0218,
      "step": 10672,
      "task_loss": 0.011321371421217918
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.045944422483444214,
      "epoch": 10.14,
      "learning_rate": 2.9120672757653916e-05,
      "loss": 0.0454,
      "step": 10673,
      "task_loss": 0.04038790985941887
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05254145711660385,
      "epoch": 10.14,
      "learning_rate": 2.9110162875002552e-05,
      "loss": 0.0481,
      "step": 10674,
      "task_loss": 0.0084177665412426
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028896521776914597,
      "epoch": 10.14,
      "learning_rate": 2.909965224572031e-05,
      "loss": 0.0289,
      "step": 10675,
      "task_loss": 0.02889561466872692
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03066609799861908,
      "epoch": 10.14,
      "learning_rate": 2.9089140871716492e-05,
      "loss": 0.0331,
      "step": 10676,
      "task_loss": 0.05538489297032356
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011451882310211658,
      "epoch": 10.14,
      "learning_rate": 2.9078628754900543e-05,
      "loss": 0.019,
      "step": 10677,
      "task_loss": 0.08714932203292847
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01488979160785675,
      "epoch": 10.14,
      "learning_rate": 2.9068115897182036e-05,
      "loss": 0.0221,
      "step": 10678,
      "task_loss": 0.08699451386928558
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.00719775166362524,
      "epoch": 10.14,
      "learning_rate": 2.905760230047068e-05,
      "loss": 0.0132,
      "step": 10679,
      "task_loss": 0.06723162531852722
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011093568056821823,
      "epoch": 10.14,
      "learning_rate": 2.9047087966676327e-05,
      "loss": 0.0141,
      "step": 10680,
      "task_loss": 0.040885455906391144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03617207705974579,
      "epoch": 10.14,
      "learning_rate": 2.903657289770896e-05,
      "loss": 0.0479,
      "step": 10681,
      "task_loss": 0.15319961309432983
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018763557076454163,
      "epoch": 10.14,
      "learning_rate": 2.902605709547868e-05,
      "loss": 0.0249,
      "step": 10682,
      "task_loss": 0.08051219582557678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016470609232783318,
      "epoch": 10.15,
      "learning_rate": 2.9015540561895738e-05,
      "loss": 0.0278,
      "step": 10683,
      "task_loss": 0.12958627939224243
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10346265137195587,
      "epoch": 10.15,
      "learning_rate": 2.9005023298870514e-05,
      "loss": 0.1048,
      "step": 10684,
      "task_loss": 0.11723846197128296
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.064060278236866,
      "epoch": 10.15,
      "learning_rate": 2.8994505308313523e-05,
      "loss": 0.0662,
      "step": 10685,
      "task_loss": 0.08514577150344849
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02188374474644661,
      "epoch": 10.15,
      "learning_rate": 2.8983986592135404e-05,
      "loss": 0.025,
      "step": 10686,
      "task_loss": 0.05310831218957901
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020624695345759392,
      "epoch": 10.15,
      "learning_rate": 2.897346715224693e-05,
      "loss": 0.0241,
      "step": 10687,
      "task_loss": 0.055748291313648224
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06058871001005173,
      "epoch": 10.15,
      "learning_rate": 2.8962946990559013e-05,
      "loss": 0.0679,
      "step": 10688,
      "task_loss": 0.13335812091827393
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009769352152943611,
      "epoch": 10.15,
      "learning_rate": 2.8952426108982693e-05,
      "loss": 0.0172,
      "step": 10689,
      "task_loss": 0.08371652662754059
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016979005187749863,
      "epoch": 10.15,
      "learning_rate": 2.8941904509429134e-05,
      "loss": 0.0216,
      "step": 10690,
      "task_loss": 0.06295529007911682
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020230475813150406,
      "epoch": 10.15,
      "learning_rate": 2.8931382193809635e-05,
      "loss": 0.0353,
      "step": 10691,
      "task_loss": 0.1709280014038086
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.041637878865003586,
      "epoch": 10.15,
      "learning_rate": 2.8920859164035625e-05,
      "loss": 0.0392,
      "step": 10692,
      "task_loss": 0.017362238839268684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04988161846995354,
      "epoch": 10.15,
      "learning_rate": 2.8910335422018664e-05,
      "loss": 0.048,
      "step": 10693,
      "task_loss": 0.031166965141892433
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021832330152392387,
      "epoch": 10.16,
      "learning_rate": 2.8899810969670448e-05,
      "loss": 0.0259,
      "step": 10694,
      "task_loss": 0.06244866922497749
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.033625852316617966,
      "epoch": 10.16,
      "learning_rate": 2.8889285808902784e-05,
      "loss": 0.0352,
      "step": 10695,
      "task_loss": 0.04985510930418968
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01881205290555954,
      "epoch": 10.16,
      "learning_rate": 2.887875994162762e-05,
      "loss": 0.0177,
      "step": 10696,
      "task_loss": 0.007423171773552895
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.031309157609939575,
      "epoch": 10.16,
      "learning_rate": 2.886823336975703e-05,
      "loss": 0.0379,
      "step": 10697,
      "task_loss": 0.09693950414657593
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02167995274066925,
      "epoch": 10.16,
      "learning_rate": 2.885770609520323e-05,
      "loss": 0.025,
      "step": 10698,
      "task_loss": 0.055332060903310776
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020276514813303947,
      "epoch": 10.16,
      "learning_rate": 2.8847178119878527e-05,
      "loss": 0.0265,
      "step": 10699,
      "task_loss": 0.08290567249059677
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029720699414610863,
      "epoch": 10.16,
      "learning_rate": 2.883664944569539e-05,
      "loss": 0.0503,
      "step": 10700,
      "task_loss": 0.2355862855911255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.037131913006305695,
      "epoch": 10.16,
      "learning_rate": 2.8826120074566414e-05,
      "loss": 0.0397,
      "step": 10701,
      "task_loss": 0.06265319883823395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02705603837966919,
      "epoch": 10.16,
      "learning_rate": 2.8815590008404293e-05,
      "loss": 0.0279,
      "step": 10702,
      "task_loss": 0.03594599664211273
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02101530134677887,
      "epoch": 10.16,
      "learning_rate": 2.8805059249121874e-05,
      "loss": 0.0222,
      "step": 10703,
      "task_loss": 0.03242805600166321
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01382765918970108,
      "epoch": 10.17,
      "learning_rate": 2.8794527798632117e-05,
      "loss": 0.023,
      "step": 10704,
      "task_loss": 0.10595635324716568
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018166912719607353,
      "epoch": 10.17,
      "learning_rate": 2.8783995658848105e-05,
      "loss": 0.023,
      "step": 10705,
      "task_loss": 0.06630286574363708
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.042599309235811234,
      "epoch": 10.17,
      "learning_rate": 2.877346283168306e-05,
      "loss": 0.0468,
      "step": 10706,
      "task_loss": 0.08451828360557556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.039667222648859024,
      "epoch": 10.17,
      "learning_rate": 2.876292931905032e-05,
      "loss": 0.0363,
      "step": 10707,
      "task_loss": 0.0058592017740011215
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.046238042414188385,
      "epoch": 10.17,
      "learning_rate": 2.875239512286335e-05,
      "loss": 0.0441,
      "step": 10708,
      "task_loss": 0.024881619960069656
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010262496769428253,
      "epoch": 10.17,
      "learning_rate": 2.8741860245035722e-05,
      "loss": 0.0095,
      "step": 10709,
      "task_loss": 0.0028820428997278214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029114559292793274,
      "epoch": 10.17,
      "learning_rate": 2.8731324687481176e-05,
      "loss": 0.0331,
      "step": 10710,
      "task_loss": 0.06866513192653656
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.047519125044345856,
      "epoch": 10.17,
      "learning_rate": 2.8720788452113517e-05,
      "loss": 0.0558,
      "step": 10711,
      "task_loss": 0.13019928336143494
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03364911675453186,
      "epoch": 10.17,
      "learning_rate": 2.8710251540846723e-05,
      "loss": 0.0376,
      "step": 10712,
      "task_loss": 0.07327281683683395
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017668629065155983,
      "epoch": 10.17,
      "learning_rate": 2.8699713955594864e-05,
      "loss": 0.019,
      "step": 10713,
      "task_loss": 0.03053228370845318
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04549794644117355,
      "epoch": 10.17,
      "learning_rate": 2.8689175698272147e-05,
      "loss": 0.0501,
      "step": 10714,
      "task_loss": 0.09190307557582855
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01674514450132847,
      "epoch": 10.18,
      "learning_rate": 2.8678636770792906e-05,
      "loss": 0.0203,
      "step": 10715,
      "task_loss": 0.05194873362779617
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018080132082104683,
      "epoch": 10.18,
      "learning_rate": 2.8668097175071572e-05,
      "loss": 0.0263,
      "step": 10716,
      "task_loss": 0.10066099464893341
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03006262332201004,
      "epoch": 10.18,
      "learning_rate": 2.865755691302272e-05,
      "loss": 0.0329,
      "step": 10717,
      "task_loss": 0.05870826542377472
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015684347599744797,
      "epoch": 10.18,
      "learning_rate": 2.864701598656104e-05,
      "loss": 0.0182,
      "step": 10718,
      "task_loss": 0.040366411209106445
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010996408760547638,
      "epoch": 10.18,
      "learning_rate": 2.8636474397601343e-05,
      "loss": 0.0106,
      "step": 10719,
      "task_loss": 0.00703669898211956
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03208538889884949,
      "epoch": 10.18,
      "learning_rate": 2.862593214805856e-05,
      "loss": 0.0318,
      "step": 10720,
      "task_loss": 0.029489625245332718
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04310718923807144,
      "epoch": 10.18,
      "learning_rate": 2.8615389239847734e-05,
      "loss": 0.0414,
      "step": 10721,
      "task_loss": 0.025636808946728706
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.061272382736206055,
      "epoch": 10.18,
      "learning_rate": 2.8604845674884045e-05,
      "loss": 0.0585,
      "step": 10722,
      "task_loss": 0.03352653980255127
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01941574551165104,
      "epoch": 10.18,
      "learning_rate": 2.8594301455082777e-05,
      "loss": 0.0198,
      "step": 10723,
      "task_loss": 0.023156698793172836
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013191182166337967,
      "epoch": 10.18,
      "learning_rate": 2.8583756582359338e-05,
      "loss": 0.0123,
      "step": 10724,
      "task_loss": 0.004337495192885399
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01237378641963005,
      "epoch": 10.19,
      "learning_rate": 2.8573211058629262e-05,
      "loss": 0.0121,
      "step": 10725,
      "task_loss": 0.009604766964912415
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020672522485256195,
      "epoch": 10.19,
      "learning_rate": 2.8562664885808176e-05,
      "loss": 0.0237,
      "step": 10726,
      "task_loss": 0.050802476704120636
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018164366483688354,
      "epoch": 10.19,
      "learning_rate": 2.8552118065811868e-05,
      "loss": 0.0254,
      "step": 10727,
      "task_loss": 0.09089714288711548
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012251758947968483,
      "epoch": 10.19,
      "learning_rate": 2.85415706005562e-05,
      "loss": 0.0149,
      "step": 10728,
      "task_loss": 0.03885126858949661
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021103456616401672,
      "epoch": 10.19,
      "learning_rate": 2.8531022491957178e-05,
      "loss": 0.0238,
      "step": 10729,
      "task_loss": 0.048550285398960114
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021369164809584618,
      "epoch": 10.19,
      "learning_rate": 2.852047374193092e-05,
      "loss": 0.0323,
      "step": 10730,
      "task_loss": 0.1303352415561676
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013989603146910667,
      "epoch": 10.19,
      "learning_rate": 2.850992435239364e-05,
      "loss": 0.0133,
      "step": 10731,
      "task_loss": 0.006928419694304466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03571196272969246,
      "epoch": 10.19,
      "learning_rate": 2.8499374325261708e-05,
      "loss": 0.0377,
      "step": 10732,
      "task_loss": 0.05529388412833214
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01848205178976059,
      "epoch": 10.19,
      "learning_rate": 2.848882366245157e-05,
      "loss": 0.023,
      "step": 10733,
      "task_loss": 0.06377825886011124
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014538046903908253,
      "epoch": 10.19,
      "learning_rate": 2.847827236587982e-05,
      "loss": 0.0203,
      "step": 10734,
      "task_loss": 0.07199336588382721
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026060445234179497,
      "epoch": 10.19,
      "learning_rate": 2.846772043746313e-05,
      "loss": 0.0239,
      "step": 10735,
      "task_loss": 0.004935260862112045
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02200356312096119,
      "epoch": 10.2,
      "learning_rate": 2.845716787911833e-05,
      "loss": 0.0247,
      "step": 10736,
      "task_loss": 0.04902653768658638
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.10178372263908386,
      "epoch": 10.2,
      "learning_rate": 2.8446614692762336e-05,
      "loss": 0.0991,
      "step": 10737,
      "task_loss": 0.07502306997776031
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03487898409366608,
      "epoch": 10.2,
      "learning_rate": 2.843606088031218e-05,
      "loss": 0.0367,
      "step": 10738,
      "task_loss": 0.05333896726369858
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014870396815240383,
      "epoch": 10.2,
      "learning_rate": 2.842550644368502e-05,
      "loss": 0.0235,
      "step": 10739,
      "task_loss": 0.1009284108877182
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017139773815870285,
      "epoch": 10.2,
      "learning_rate": 2.841495138479811e-05,
      "loss": 0.0162,
      "step": 10740,
      "task_loss": 0.008153453469276428
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02908232994377613,
      "epoch": 10.2,
      "learning_rate": 2.8404395705568848e-05,
      "loss": 0.0271,
      "step": 10741,
      "task_loss": 0.008917009457945824
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04850727319717407,
      "epoch": 10.2,
      "learning_rate": 2.8393839407914702e-05,
      "loss": 0.0502,
      "step": 10742,
      "task_loss": 0.06578432023525238
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017160603776574135,
      "epoch": 10.2,
      "learning_rate": 2.8383282493753283e-05,
      "loss": 0.0163,
      "step": 10743,
      "task_loss": 0.008876651525497437
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03274647891521454,
      "epoch": 10.2,
      "learning_rate": 2.83727249650023e-05,
      "loss": 0.0303,
      "step": 10744,
      "task_loss": 0.008528593927621841
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023494988679885864,
      "epoch": 10.2,
      "learning_rate": 2.836216682357959e-05,
      "loss": 0.0296,
      "step": 10745,
      "task_loss": 0.08411206305027008
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.045075200498104095,
      "epoch": 10.21,
      "learning_rate": 2.8351608071403085e-05,
      "loss": 0.0411,
      "step": 10746,
      "task_loss": 0.005580326542258263
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07080404460430145,
      "epoch": 10.21,
      "learning_rate": 2.8341048710390832e-05,
      "loss": 0.0653,
      "step": 10747,
      "task_loss": 0.016258470714092255
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02719147875905037,
      "epoch": 10.21,
      "learning_rate": 2.8330488742460987e-05,
      "loss": 0.0254,
      "step": 10748,
      "task_loss": 0.009210776537656784
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03383823484182358,
      "epoch": 10.21,
      "learning_rate": 2.8319928169531825e-05,
      "loss": 0.0556,
      "step": 10749,
      "task_loss": 0.2511603534221649
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014685697853565216,
      "epoch": 10.21,
      "learning_rate": 2.830936699352172e-05,
      "loss": 0.0209,
      "step": 10750,
      "task_loss": 0.07651330530643463
    },
    {
      "epoch": 10.21,
      "eval_accuracy": 0.9025229357798165,
      "eval_loss": 0.4256412982940674,
      "eval_runtime": 18.2731,
      "eval_samples_per_second": 47.721,
      "eval_steps_per_second": 5.965,
      "step": 10750
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014020141214132309,
      "epoch": 10.21,
      "learning_rate": 2.8298805216349167e-05,
      "loss": 0.021,
      "step": 10751,
      "task_loss": 0.08416274189949036
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03295741230249405,
      "epoch": 10.21,
      "learning_rate": 2.8288242839932744e-05,
      "loss": 0.0485,
      "step": 10752,
      "task_loss": 0.18834072351455688
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013736523687839508,
      "epoch": 10.21,
      "learning_rate": 2.8277679866191194e-05,
      "loss": 0.0133,
      "step": 10753,
      "task_loss": 0.009392468258738518
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05337664112448692,
      "epoch": 10.21,
      "learning_rate": 2.8267116297043294e-05,
      "loss": 0.0524,
      "step": 10754,
      "task_loss": 0.0431857630610466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06362185627222061,
      "epoch": 10.21,
      "learning_rate": 2.8256552134407993e-05,
      "loss": 0.0686,
      "step": 10755,
      "task_loss": 0.11329066753387451
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.050114959478378296,
      "epoch": 10.21,
      "learning_rate": 2.8245987380204313e-05,
      "loss": 0.0566,
      "step": 10756,
      "task_loss": 0.11529053747653961
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03267570957541466,
      "epoch": 10.22,
      "learning_rate": 2.8235422036351382e-05,
      "loss": 0.0373,
      "step": 10757,
      "task_loss": 0.07926565408706665
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013439756818115711,
      "epoch": 10.22,
      "learning_rate": 2.822485610476847e-05,
      "loss": 0.013,
      "step": 10758,
      "task_loss": 0.008762186393141747
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02206001617014408,
      "epoch": 10.22,
      "learning_rate": 2.8214289587374908e-05,
      "loss": 0.0288,
      "step": 10759,
      "task_loss": 0.08915533125400543
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01932983100414276,
      "epoch": 10.22,
      "learning_rate": 2.8203722486090168e-05,
      "loss": 0.0201,
      "step": 10760,
      "task_loss": 0.026729634031653404
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012917459942400455,
      "epoch": 10.22,
      "learning_rate": 2.8193154802833803e-05,
      "loss": 0.0127,
      "step": 10761,
      "task_loss": 0.010323688387870789
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022919047623872757,
      "epoch": 10.22,
      "learning_rate": 2.818258653952549e-05,
      "loss": 0.0217,
      "step": 10762,
      "task_loss": 0.01115039736032486
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01916324719786644,
      "epoch": 10.22,
      "learning_rate": 2.8172017698085013e-05,
      "loss": 0.018,
      "step": 10763,
      "task_loss": 0.00737486407160759
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01026608981192112,
      "epoch": 10.22,
      "learning_rate": 2.816144828043224e-05,
      "loss": 0.0163,
      "step": 10764,
      "task_loss": 0.07105138152837753
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018040932714939117,
      "epoch": 10.22,
      "learning_rate": 2.8150878288487155e-05,
      "loss": 0.0168,
      "step": 10765,
      "task_loss": 0.0060977693647146225
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012006964534521103,
      "epoch": 10.22,
      "learning_rate": 2.8140307724169857e-05,
      "loss": 0.0113,
      "step": 10766,
      "task_loss": 0.004458732903003693
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.036806024610996246,
      "epoch": 10.23,
      "learning_rate": 2.812973658940054e-05,
      "loss": 0.0391,
      "step": 10767,
      "task_loss": 0.05974595993757248
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.039469681680202484,
      "epoch": 10.23,
      "learning_rate": 2.8119164886099504e-05,
      "loss": 0.048,
      "step": 10768,
      "task_loss": 0.12487009167671204
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011058198288083076,
      "epoch": 10.23,
      "learning_rate": 2.8108592616187133e-05,
      "loss": 0.0105,
      "step": 10769,
      "task_loss": 0.005702011287212372
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01453112531453371,
      "epoch": 10.23,
      "learning_rate": 2.8098019781583944e-05,
      "loss": 0.0201,
      "step": 10770,
      "task_loss": 0.07058731466531754
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01812460459768772,
      "epoch": 10.23,
      "learning_rate": 2.8087446384210547e-05,
      "loss": 0.0173,
      "step": 10771,
      "task_loss": 0.01019604504108429
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014030968770384789,
      "epoch": 10.23,
      "learning_rate": 2.8076872425987637e-05,
      "loss": 0.0138,
      "step": 10772,
      "task_loss": 0.011378861963748932
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02499603107571602,
      "epoch": 10.23,
      "learning_rate": 2.8066297908836043e-05,
      "loss": 0.0273,
      "step": 10773,
      "task_loss": 0.04773323982954025
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04917794466018677,
      "epoch": 10.23,
      "learning_rate": 2.8055722834676658e-05,
      "loss": 0.0502,
      "step": 10774,
      "task_loss": 0.05955827608704567
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04151075705885887,
      "epoch": 10.23,
      "learning_rate": 2.804514720543051e-05,
      "loss": 0.0442,
      "step": 10775,
      "task_loss": 0.06856313347816467
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06127103045582771,
      "epoch": 10.23,
      "learning_rate": 2.80345710230187e-05,
      "loss": 0.0622,
      "step": 10776,
      "task_loss": 0.07090768218040466
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015070604160428047,
      "epoch": 10.23,
      "learning_rate": 2.802399428936246e-05,
      "loss": 0.0141,
      "step": 10777,
      "task_loss": 0.004883896559476852
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017283853143453598,
      "epoch": 10.24,
      "learning_rate": 2.8013417006383076e-05,
      "loss": 0.0266,
      "step": 10778,
      "task_loss": 0.11040695756673813
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018551714718341827,
      "epoch": 10.24,
      "learning_rate": 2.8002839176001987e-05,
      "loss": 0.0171,
      "step": 10779,
      "task_loss": 0.004415277391672134
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02501179277896881,
      "epoch": 10.24,
      "learning_rate": 2.799226080014071e-05,
      "loss": 0.0422,
      "step": 10780,
      "task_loss": 0.19670268893241882
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012673512101173401,
      "epoch": 10.24,
      "learning_rate": 2.7981681880720838e-05,
      "loss": 0.0133,
      "step": 10781,
      "task_loss": 0.018554171547293663
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014598320238292217,
      "epoch": 10.24,
      "learning_rate": 2.7971102419664103e-05,
      "loss": 0.0141,
      "step": 10782,
      "task_loss": 0.009689368307590485
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014966873452067375,
      "epoch": 10.24,
      "learning_rate": 2.7960522418892288e-05,
      "loss": 0.0262,
      "step": 10783,
      "task_loss": 0.12777450680732727
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01800619065761566,
      "epoch": 10.24,
      "learning_rate": 2.794994188032733e-05,
      "loss": 0.0173,
      "step": 10784,
      "task_loss": 0.011067090556025505
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01135667972266674,
      "epoch": 10.24,
      "learning_rate": 2.7939360805891218e-05,
      "loss": 0.0107,
      "step": 10785,
      "task_loss": 0.004777856171131134
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025919828563928604,
      "epoch": 10.24,
      "learning_rate": 2.7928779197506056e-05,
      "loss": 0.0289,
      "step": 10786,
      "task_loss": 0.05572760850191116
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016921542584896088,
      "epoch": 10.24,
      "learning_rate": 2.7918197057094054e-05,
      "loss": 0.0157,
      "step": 10787,
      "task_loss": 0.004976712167263031
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025802042335271835,
      "epoch": 10.25,
      "learning_rate": 2.7907614386577497e-05,
      "loss": 0.0428,
      "step": 10788,
      "task_loss": 0.19616416096687317
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008495388552546501,
      "epoch": 10.25,
      "learning_rate": 2.789703118787879e-05,
      "loss": 0.0079,
      "step": 10789,
      "task_loss": 0.002873443067073822
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016510333865880966,
      "epoch": 10.25,
      "learning_rate": 2.7886447462920412e-05,
      "loss": 0.0287,
      "step": 10790,
      "task_loss": 0.13880857825279236
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03891441598534584,
      "epoch": 10.25,
      "learning_rate": 2.787586321362495e-05,
      "loss": 0.046,
      "step": 10791,
      "task_loss": 0.1098531037569046
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012705841101706028,
      "epoch": 10.25,
      "learning_rate": 2.7865278441915082e-05,
      "loss": 0.0149,
      "step": 10792,
      "task_loss": 0.034181784838438034
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01141888927668333,
      "epoch": 10.25,
      "learning_rate": 2.785469314971359e-05,
      "loss": 0.0187,
      "step": 10793,
      "task_loss": 0.08427157998085022
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017081687226891518,
      "epoch": 10.25,
      "learning_rate": 2.7844107338943343e-05,
      "loss": 0.0201,
      "step": 10794,
      "task_loss": 0.0467962883412838
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009645327925682068,
      "epoch": 10.25,
      "learning_rate": 2.7833521011527293e-05,
      "loss": 0.0128,
      "step": 10795,
      "task_loss": 0.041607990860939026
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.032230883836746216,
      "epoch": 10.25,
      "learning_rate": 2.782293416938851e-05,
      "loss": 0.0446,
      "step": 10796,
      "task_loss": 0.15550780296325684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07105334848165512,
      "epoch": 10.25,
      "learning_rate": 2.7812346814450135e-05,
      "loss": 0.0718,
      "step": 10797,
      "task_loss": 0.07855713367462158
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013354838825762272,
      "epoch": 10.25,
      "learning_rate": 2.7801758948635414e-05,
      "loss": 0.0164,
      "step": 10798,
      "task_loss": 0.04363374784588814
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009988810867071152,
      "epoch": 10.26,
      "learning_rate": 2.77911705738677e-05,
      "loss": 0.0175,
      "step": 10799,
      "task_loss": 0.08463583886623383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015538221225142479,
      "epoch": 10.26,
      "learning_rate": 2.7780581692070395e-05,
      "loss": 0.0146,
      "step": 10800,
      "task_loss": 0.006343167275190353
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01280184555798769,
      "epoch": 10.26,
      "learning_rate": 2.7769992305167043e-05,
      "loss": 0.0203,
      "step": 10801,
      "task_loss": 0.08731023967266083
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05490295588970184,
      "epoch": 10.26,
      "learning_rate": 2.775940241508124e-05,
      "loss": 0.0581,
      "step": 10802,
      "task_loss": 0.08693551272153854
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.006106989458203316,
      "epoch": 10.26,
      "learning_rate": 2.774881202373671e-05,
      "loss": 0.0136,
      "step": 10803,
      "task_loss": 0.0810597613453865
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05367905646562576,
      "epoch": 10.26,
      "learning_rate": 2.773822113305723e-05,
      "loss": 0.0661,
      "step": 10804,
      "task_loss": 0.17820340394973755
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011592602357268333,
      "epoch": 10.26,
      "learning_rate": 2.7727629744966695e-05,
      "loss": 0.0109,
      "step": 10805,
      "task_loss": 0.004181232303380966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016554832458496094,
      "epoch": 10.26,
      "learning_rate": 2.7717037861389082e-05,
      "loss": 0.016,
      "step": 10806,
      "task_loss": 0.010724140331149101
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015016058459877968,
      "epoch": 10.26,
      "learning_rate": 2.7706445484248454e-05,
      "loss": 0.0143,
      "step": 10807,
      "task_loss": 0.007696835324168205
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.00887630321085453,
      "epoch": 10.26,
      "learning_rate": 2.769585261546897e-05,
      "loss": 0.013,
      "step": 10808,
      "task_loss": 0.05019646883010864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02640136145055294,
      "epoch": 10.26,
      "learning_rate": 2.768525925697487e-05,
      "loss": 0.0284,
      "step": 10809,
      "task_loss": 0.04624557122588158
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.08202344924211502,
      "epoch": 10.27,
      "learning_rate": 2.76746654106905e-05,
      "loss": 0.0864,
      "step": 10810,
      "task_loss": 0.125411257147789
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.051587365567684174,
      "epoch": 10.27,
      "learning_rate": 2.7664071078540282e-05,
      "loss": 0.0557,
      "step": 10811,
      "task_loss": 0.0922786295413971
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01615780219435692,
      "epoch": 10.27,
      "learning_rate": 2.7653476262448713e-05,
      "loss": 0.0155,
      "step": 10812,
      "task_loss": 0.009321728721261024
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06020238250494003,
      "epoch": 10.27,
      "learning_rate": 2.76428809643404e-05,
      "loss": 0.0655,
      "step": 10813,
      "task_loss": 0.11301226913928986
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025619039312005043,
      "epoch": 10.27,
      "learning_rate": 2.763228518614004e-05,
      "loss": 0.0258,
      "step": 10814,
      "task_loss": 0.027688482776284218
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011568024754524231,
      "epoch": 10.27,
      "learning_rate": 2.7621688929772393e-05,
      "loss": 0.0181,
      "step": 10815,
      "task_loss": 0.07710108906030655
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.048748280853033066,
      "epoch": 10.27,
      "learning_rate": 2.761109219716233e-05,
      "loss": 0.0455,
      "step": 10816,
      "task_loss": 0.016586463898420334
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03522196784615517,
      "epoch": 10.27,
      "learning_rate": 2.760049499023479e-05,
      "loss": 0.0463,
      "step": 10817,
      "task_loss": 0.1458013653755188
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010583333671092987,
      "epoch": 10.27,
      "learning_rate": 2.7589897310914814e-05,
      "loss": 0.0098,
      "step": 10818,
      "task_loss": 0.002439044415950775
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0179889015853405,
      "epoch": 10.27,
      "learning_rate": 2.7579299161127513e-05,
      "loss": 0.0167,
      "step": 10819,
      "task_loss": 0.0054456982761621475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014011922292411327,
      "epoch": 10.28,
      "learning_rate": 2.756870054279811e-05,
      "loss": 0.0139,
      "step": 10820,
      "task_loss": 0.01266825757920742
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020268365740776062,
      "epoch": 10.28,
      "learning_rate": 2.755810145785187e-05,
      "loss": 0.0187,
      "step": 10821,
      "task_loss": 0.004288617521524429
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025478918105363846,
      "epoch": 10.28,
      "learning_rate": 2.754750190821418e-05,
      "loss": 0.025,
      "step": 10822,
      "task_loss": 0.020617567002773285
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021658936515450478,
      "epoch": 10.28,
      "learning_rate": 2.753690189581051e-05,
      "loss": 0.0199,
      "step": 10823,
      "task_loss": 0.004049813374876976
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06283696740865707,
      "epoch": 10.28,
      "learning_rate": 2.752630142256638e-05,
      "loss": 0.0597,
      "step": 10824,
      "task_loss": 0.03175070881843567
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011753297410905361,
      "epoch": 10.28,
      "learning_rate": 2.7515700490407443e-05,
      "loss": 0.0111,
      "step": 10825,
      "task_loss": 0.00544341653585434
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09492142498493195,
      "epoch": 10.28,
      "learning_rate": 2.7505099101259386e-05,
      "loss": 0.0944,
      "step": 10826,
      "task_loss": 0.08998773992061615
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.007941950112581253,
      "epoch": 10.28,
      "learning_rate": 2.749449725704802e-05,
      "loss": 0.0077,
      "step": 10827,
      "task_loss": 0.005958493798971176
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018316073343157768,
      "epoch": 10.28,
      "learning_rate": 2.748389495969921e-05,
      "loss": 0.0302,
      "step": 10828,
      "task_loss": 0.1371515393257141
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022628696635365486,
      "epoch": 10.28,
      "learning_rate": 2.747329221113891e-05,
      "loss": 0.0268,
      "step": 10829,
      "task_loss": 0.06442567706108093
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012505222111940384,
      "epoch": 10.28,
      "learning_rate": 2.7462689013293176e-05,
      "loss": 0.0118,
      "step": 10830,
      "task_loss": 0.005520684644579887
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02343965508043766,
      "epoch": 10.29,
      "learning_rate": 2.745208536808812e-05,
      "loss": 0.0239,
      "step": 10831,
      "task_loss": 0.027552150189876556
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05450985208153725,
      "epoch": 10.29,
      "learning_rate": 2.7441481277449954e-05,
      "loss": 0.06,
      "step": 10832,
      "task_loss": 0.10902714729309082
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023538943380117416,
      "epoch": 10.29,
      "learning_rate": 2.743087674330495e-05,
      "loss": 0.0287,
      "step": 10833,
      "task_loss": 0.07561871409416199
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01717299409210682,
      "epoch": 10.29,
      "learning_rate": 2.742027176757948e-05,
      "loss": 0.0172,
      "step": 10834,
      "task_loss": 0.01727975904941559
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02774146758019924,
      "epoch": 10.29,
      "learning_rate": 2.7409666352199986e-05,
      "loss": 0.0257,
      "step": 10835,
      "task_loss": 0.007454710081219673
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020189553499221802,
      "epoch": 10.29,
      "learning_rate": 2.7399060499092992e-05,
      "loss": 0.0191,
      "step": 10836,
      "task_loss": 0.009731443598866463
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06831050664186478,
      "epoch": 10.29,
      "learning_rate": 2.7388454210185115e-05,
      "loss": 0.067,
      "step": 10837,
      "task_loss": 0.055503882467746735
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013829356990754604,
      "epoch": 10.29,
      "learning_rate": 2.7377847487403018e-05,
      "loss": 0.0128,
      "step": 10838,
      "task_loss": 0.003266597166657448
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014532563276588917,
      "epoch": 10.29,
      "learning_rate": 2.736724033267347e-05,
      "loss": 0.0153,
      "step": 10839,
      "task_loss": 0.021880408748984337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05628987401723862,
      "epoch": 10.29,
      "learning_rate": 2.7356632747923322e-05,
      "loss": 0.0626,
      "step": 10840,
      "task_loss": 0.11941660940647125
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.051775701344013214,
      "epoch": 10.3,
      "learning_rate": 2.7346024735079486e-05,
      "loss": 0.0678,
      "step": 10841,
      "task_loss": 0.21175605058670044
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027265068143606186,
      "epoch": 10.3,
      "learning_rate": 2.7335416296068962e-05,
      "loss": 0.0256,
      "step": 10842,
      "task_loss": 0.010892918333411217
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028751587495207787,
      "epoch": 10.3,
      "learning_rate": 2.7324807432818805e-05,
      "loss": 0.0302,
      "step": 10843,
      "task_loss": 0.043142762035131454
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01213914155960083,
      "epoch": 10.3,
      "learning_rate": 2.731419814725619e-05,
      "loss": 0.0111,
      "step": 10844,
      "task_loss": 0.002093670889735222
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021539079025387764,
      "epoch": 10.3,
      "learning_rate": 2.730358844130834e-05,
      "loss": 0.0212,
      "step": 10845,
      "task_loss": 0.018369406461715698
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03298931196331978,
      "epoch": 10.3,
      "learning_rate": 2.729297831690255e-05,
      "loss": 0.0398,
      "step": 10846,
      "task_loss": 0.10142332315444946
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016631023958325386,
      "epoch": 10.3,
      "learning_rate": 2.728236777596621e-05,
      "loss": 0.017,
      "step": 10847,
      "task_loss": 0.020175570622086525
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011871835216879845,
      "epoch": 10.3,
      "learning_rate": 2.7271756820426763e-05,
      "loss": 0.0125,
      "step": 10848,
      "task_loss": 0.017757927998900414
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0549296960234642,
      "epoch": 10.3,
      "learning_rate": 2.7261145452211763e-05,
      "loss": 0.0602,
      "step": 10849,
      "task_loss": 0.10732554644346237
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020789777860045433,
      "epoch": 10.3,
      "learning_rate": 2.725053367324879e-05,
      "loss": 0.0202,
      "step": 10850,
      "task_loss": 0.014408687129616737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011434385553002357,
      "epoch": 10.3,
      "learning_rate": 2.723992148546554e-05,
      "loss": 0.0134,
      "step": 10851,
      "task_loss": 0.031306661665439606
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014954166486859322,
      "epoch": 10.31,
      "learning_rate": 2.7229308890789767e-05,
      "loss": 0.0181,
      "step": 10852,
      "task_loss": 0.046106450259685516
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010979441925883293,
      "epoch": 10.31,
      "learning_rate": 2.7218695891149293e-05,
      "loss": 0.0102,
      "step": 10853,
      "task_loss": 0.003026876598596573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015950333327054977,
      "epoch": 10.31,
      "learning_rate": 2.720808248847203e-05,
      "loss": 0.0345,
      "step": 10854,
      "task_loss": 0.2015371024608612
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02772637829184532,
      "epoch": 10.31,
      "learning_rate": 2.719746868468595e-05,
      "loss": 0.0369,
      "step": 10855,
      "task_loss": 0.11981463432312012
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.031578004360198975,
      "epoch": 10.31,
      "learning_rate": 2.7186854481719092e-05,
      "loss": 0.0397,
      "step": 10856,
      "task_loss": 0.11244003474712372
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.007710220292210579,
      "epoch": 10.31,
      "learning_rate": 2.7176239881499595e-05,
      "loss": 0.0072,
      "step": 10857,
      "task_loss": 0.002641640603542328
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009416783228516579,
      "epoch": 10.31,
      "learning_rate": 2.716562488595563e-05,
      "loss": 0.0152,
      "step": 10858,
      "task_loss": 0.06755001842975616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06231432408094406,
      "epoch": 10.31,
      "learning_rate": 2.715500949701549e-05,
      "loss": 0.0701,
      "step": 10859,
      "task_loss": 0.14002208411693573
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027409039437770844,
      "epoch": 10.31,
      "learning_rate": 2.7144393716607486e-05,
      "loss": 0.0394,
      "step": 10860,
      "task_loss": 0.147533118724823
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04957464337348938,
      "epoch": 10.31,
      "learning_rate": 2.713377754666004e-05,
      "loss": 0.0626,
      "step": 10861,
      "task_loss": 0.17979982495307922
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05504240095615387,
      "epoch": 10.32,
      "learning_rate": 2.712316098910162e-05,
      "loss": 0.0618,
      "step": 10862,
      "task_loss": 0.12275400012731552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015080356039106846,
      "epoch": 10.32,
      "learning_rate": 2.711254404586079e-05,
      "loss": 0.0287,
      "step": 10863,
      "task_loss": 0.1511533409357071
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008478467352688313,
      "epoch": 10.32,
      "learning_rate": 2.7101926718866156e-05,
      "loss": 0.014,
      "step": 10864,
      "task_loss": 0.06384154409170151
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.1419282853603363,
      "epoch": 10.32,
      "learning_rate": 2.7091309010046408e-05,
      "loss": 0.1438,
      "step": 10865,
      "task_loss": 0.1609189212322235
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018446022644639015,
      "epoch": 10.32,
      "learning_rate": 2.708069092133031e-05,
      "loss": 0.0171,
      "step": 10866,
      "task_loss": 0.005260376259684563
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0161033496260643,
      "epoch": 10.32,
      "learning_rate": 2.7070072454646683e-05,
      "loss": 0.0208,
      "step": 10867,
      "task_loss": 0.06297904253005981
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06171935796737671,
      "epoch": 10.32,
      "learning_rate": 2.7059453611924433e-05,
      "loss": 0.0676,
      "step": 10868,
      "task_loss": 0.12061215192079544
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.028885137289762497,
      "epoch": 10.32,
      "learning_rate": 2.7048834395092505e-05,
      "loss": 0.0334,
      "step": 10869,
      "task_loss": 0.07451960444450378
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011872227303683758,
      "epoch": 10.32,
      "learning_rate": 2.7038214806079948e-05,
      "loss": 0.0111,
      "step": 10870,
      "task_loss": 0.0037200450897216797
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03222740814089775,
      "epoch": 10.32,
      "learning_rate": 2.702759484681585e-05,
      "loss": 0.0405,
      "step": 10871,
      "task_loss": 0.11447662115097046
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013929888606071472,
      "epoch": 10.32,
      "learning_rate": 2.701697451922939e-05,
      "loss": 0.0206,
      "step": 10872,
      "task_loss": 0.08043880015611649
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013857328332960606,
      "epoch": 10.33,
      "learning_rate": 2.7006353825249792e-05,
      "loss": 0.015,
      "step": 10873,
      "task_loss": 0.025718068704009056
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020384974777698517,
      "epoch": 10.33,
      "learning_rate": 2.6995732766806354e-05,
      "loss": 0.0243,
      "step": 10874,
      "task_loss": 0.059511590749025345
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010584133677184582,
      "epoch": 10.33,
      "learning_rate": 2.6985111345828452e-05,
      "loss": 0.0111,
      "step": 10875,
      "task_loss": 0.01563730277121067
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03728438913822174,
      "epoch": 10.33,
      "learning_rate": 2.6974489564245513e-05,
      "loss": 0.038,
      "step": 10876,
      "task_loss": 0.044412482529878616
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018023356795310974,
      "epoch": 10.33,
      "learning_rate": 2.6963867423987032e-05,
      "loss": 0.0171,
      "step": 10877,
      "task_loss": 0.008735572919249535
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024205505847930908,
      "epoch": 10.33,
      "learning_rate": 2.695324492698258e-05,
      "loss": 0.0222,
      "step": 10878,
      "task_loss": 0.004576884210109711
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04197827726602554,
      "epoch": 10.33,
      "learning_rate": 2.694262207516178e-05,
      "loss": 0.0406,
      "step": 10879,
      "task_loss": 0.028400206938385963
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015484409406781197,
      "epoch": 10.33,
      "learning_rate": 2.6931998870454327e-05,
      "loss": 0.0145,
      "step": 10880,
      "task_loss": 0.005911100655794144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014505671337246895,
      "epoch": 10.33,
      "learning_rate": 2.692137531478997e-05,
      "loss": 0.0135,
      "step": 10881,
      "task_loss": 0.003982797265052795
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024028435349464417,
      "epoch": 10.33,
      "learning_rate": 2.6910751410098532e-05,
      "loss": 0.0294,
      "step": 10882,
      "task_loss": 0.07786644250154495
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05532676726579666,
      "epoch": 10.34,
      "learning_rate": 2.6900127158309903e-05,
      "loss": 0.0686,
      "step": 10883,
      "task_loss": 0.18820956349372864
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.00881568156182766,
      "epoch": 10.34,
      "learning_rate": 2.688950256135402e-05,
      "loss": 0.0155,
      "step": 10884,
      "task_loss": 0.07542967796325684
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.007144515868276358,
      "epoch": 10.34,
      "learning_rate": 2.6878877621160904e-05,
      "loss": 0.0068,
      "step": 10885,
      "task_loss": 0.0034852921962738037
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03914850950241089,
      "epoch": 10.34,
      "learning_rate": 2.686825233966061e-05,
      "loss": 0.036,
      "step": 10886,
      "task_loss": 0.007247054949402809
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03601888567209244,
      "epoch": 10.34,
      "learning_rate": 2.6857626718783285e-05,
      "loss": 0.0516,
      "step": 10887,
      "task_loss": 0.19196146726608276
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0399981364607811,
      "epoch": 10.34,
      "learning_rate": 2.6847000760459118e-05,
      "loss": 0.0487,
      "step": 10888,
      "task_loss": 0.12683525681495667
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.007798383943736553,
      "epoch": 10.34,
      "learning_rate": 2.683637446661837e-05,
      "loss": 0.0138,
      "step": 10889,
      "task_loss": 0.06797172874212265
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04516351968050003,
      "epoch": 10.34,
      "learning_rate": 2.6825747839191362e-05,
      "loss": 0.0543,
      "step": 10890,
      "task_loss": 0.13696971535682678
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05506161227822304,
      "epoch": 10.34,
      "learning_rate": 2.681512088010845e-05,
      "loss": 0.0671,
      "step": 10891,
      "task_loss": 0.1754119098186493
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.034937888383865356,
      "epoch": 10.34,
      "learning_rate": 2.6804493591300105e-05,
      "loss": 0.0372,
      "step": 10892,
      "task_loss": 0.05717271566390991
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01608917862176895,
      "epoch": 10.34,
      "learning_rate": 2.6793865974696803e-05,
      "loss": 0.0375,
      "step": 10893,
      "task_loss": 0.23010924458503723
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0070289126597344875,
      "epoch": 10.35,
      "learning_rate": 2.67832380322291e-05,
      "loss": 0.0152,
      "step": 10894,
      "task_loss": 0.0884644091129303
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02088269032537937,
      "epoch": 10.35,
      "learning_rate": 2.6772609765827627e-05,
      "loss": 0.0205,
      "step": 10895,
      "task_loss": 0.016729770228266716
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021521752700209618,
      "epoch": 10.35,
      "learning_rate": 2.6761981177423052e-05,
      "loss": 0.0341,
      "step": 10896,
      "task_loss": 0.14689095318317413
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.030128255486488342,
      "epoch": 10.35,
      "learning_rate": 2.6751352268946118e-05,
      "loss": 0.0355,
      "step": 10897,
      "task_loss": 0.08407264947891235
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01246646698564291,
      "epoch": 10.35,
      "learning_rate": 2.6740723042327598e-05,
      "loss": 0.0171,
      "step": 10898,
      "task_loss": 0.0584290474653244
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01876392588019371,
      "epoch": 10.35,
      "learning_rate": 2.673009349949836e-05,
      "loss": 0.0292,
      "step": 10899,
      "task_loss": 0.12266287952661514
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03964090347290039,
      "epoch": 10.35,
      "learning_rate": 2.6719463642389302e-05,
      "loss": 0.0393,
      "step": 10900,
      "task_loss": 0.036599088460206985
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020488981157541275,
      "epoch": 10.35,
      "learning_rate": 2.6708833472931394e-05,
      "loss": 0.0342,
      "step": 10901,
      "task_loss": 0.15787988901138306
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07471579313278198,
      "epoch": 10.35,
      "learning_rate": 2.669820299305566e-05,
      "loss": 0.0884,
      "step": 10902,
      "task_loss": 0.21201379597187042
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.027081774547696114,
      "epoch": 10.35,
      "learning_rate": 2.6687572204693174e-05,
      "loss": 0.0261,
      "step": 10903,
      "task_loss": 0.017036577686667442
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026169799268245697,
      "epoch": 10.36,
      "learning_rate": 2.667694110977506e-05,
      "loss": 0.0241,
      "step": 10904,
      "task_loss": 0.005672993138432503
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020091958343982697,
      "epoch": 10.36,
      "learning_rate": 2.6666309710232522e-05,
      "loss": 0.0294,
      "step": 10905,
      "task_loss": 0.11267304420471191
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.006835642736405134,
      "epoch": 10.36,
      "learning_rate": 2.6655678007996804e-05,
      "loss": 0.0068,
      "step": 10906,
      "task_loss": 0.006012851372361183
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05445358157157898,
      "epoch": 10.36,
      "learning_rate": 2.66450460049992e-05,
      "loss": 0.0629,
      "step": 10907,
      "task_loss": 0.13931649923324585
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024199290201067924,
      "epoch": 10.36,
      "learning_rate": 2.6634413703171058e-05,
      "loss": 0.0294,
      "step": 10908,
      "task_loss": 0.07624688744544983
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014362799003720284,
      "epoch": 10.36,
      "learning_rate": 2.6623781104443806e-05,
      "loss": 0.0138,
      "step": 10909,
      "task_loss": 0.009138602763414383
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.029530849307775497,
      "epoch": 10.36,
      "learning_rate": 2.6613148210748894e-05,
      "loss": 0.0279,
      "step": 10910,
      "task_loss": 0.01313771866261959
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04332811385393143,
      "epoch": 10.36,
      "learning_rate": 2.6602515024017842e-05,
      "loss": 0.045,
      "step": 10911,
      "task_loss": 0.06038268655538559
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0265837199985981,
      "epoch": 10.36,
      "learning_rate": 2.6591881546182216e-05,
      "loss": 0.0263,
      "step": 10912,
      "task_loss": 0.024174809455871582
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023015733808279037,
      "epoch": 10.36,
      "learning_rate": 2.6581247779173635e-05,
      "loss": 0.0233,
      "step": 10913,
      "task_loss": 0.026104921475052834
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02378995530307293,
      "epoch": 10.36,
      "learning_rate": 2.6570613724923788e-05,
      "loss": 0.0292,
      "step": 10914,
      "task_loss": 0.07820422202348709
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011787974275648594,
      "epoch": 10.37,
      "learning_rate": 2.655997938536439e-05,
      "loss": 0.0193,
      "step": 10915,
      "task_loss": 0.08679046481847763
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023985158652067184,
      "epoch": 10.37,
      "learning_rate": 2.654934476242723e-05,
      "loss": 0.0221,
      "step": 10916,
      "task_loss": 0.005446845665574074
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.025475893169641495,
      "epoch": 10.37,
      "learning_rate": 2.653870985804412e-05,
      "loss": 0.024,
      "step": 10917,
      "task_loss": 0.010533835738897324
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04181814193725586,
      "epoch": 10.37,
      "learning_rate": 2.6528074674146963e-05,
      "loss": 0.052,
      "step": 10918,
      "task_loss": 0.14384526014328003
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020111829042434692,
      "epoch": 10.37,
      "learning_rate": 2.6517439212667677e-05,
      "loss": 0.0193,
      "step": 10919,
      "task_loss": 0.01181393675506115
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016157563775777817,
      "epoch": 10.37,
      "learning_rate": 2.6506803475538256e-05,
      "loss": 0.0182,
      "step": 10920,
      "task_loss": 0.03620835393667221
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026654712855815887,
      "epoch": 10.37,
      "learning_rate": 2.649616746469072e-05,
      "loss": 0.0281,
      "step": 10921,
      "task_loss": 0.04074002057313919
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014352642931044102,
      "epoch": 10.37,
      "learning_rate": 2.648553118205716e-05,
      "loss": 0.0138,
      "step": 10922,
      "task_loss": 0.009018674492835999
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02536664716899395,
      "epoch": 10.37,
      "learning_rate": 2.6474894629569713e-05,
      "loss": 0.0348,
      "step": 10923,
      "task_loss": 0.11997328698635101
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016859643161296844,
      "epoch": 10.37,
      "learning_rate": 2.6464257809160548e-05,
      "loss": 0.023,
      "step": 10924,
      "task_loss": 0.07873199135065079
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01845327578485012,
      "epoch": 10.38,
      "learning_rate": 2.6453620722761896e-05,
      "loss": 0.022,
      "step": 10925,
      "task_loss": 0.05434544384479523
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008304309099912643,
      "epoch": 10.38,
      "learning_rate": 2.6442983372306045e-05,
      "loss": 0.0079,
      "step": 10926,
      "task_loss": 0.004198441281914711
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021259469911456108,
      "epoch": 10.38,
      "learning_rate": 2.643234575972531e-05,
      "loss": 0.0265,
      "step": 10927,
      "task_loss": 0.07323883473873138
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008786037564277649,
      "epoch": 10.38,
      "learning_rate": 2.642170788695208e-05,
      "loss": 0.0108,
      "step": 10928,
      "task_loss": 0.02885708585381508
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04073190689086914,
      "epoch": 10.38,
      "learning_rate": 2.6411069755918755e-05,
      "loss": 0.044,
      "step": 10929,
      "task_loss": 0.07315443456172943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02872268483042717,
      "epoch": 10.38,
      "learning_rate": 2.6400431368557815e-05,
      "loss": 0.035,
      "step": 10930,
      "task_loss": 0.09104539453983307
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.007937128655612469,
      "epoch": 10.38,
      "learning_rate": 2.6389792726801778e-05,
      "loss": 0.0074,
      "step": 10931,
      "task_loss": 0.0028884951025247574
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04766330122947693,
      "epoch": 10.38,
      "learning_rate": 2.6379153832583186e-05,
      "loss": 0.0664,
      "step": 10932,
      "task_loss": 0.23463010787963867
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03247709199786186,
      "epoch": 10.38,
      "learning_rate": 2.6368514687834672e-05,
      "loss": 0.0347,
      "step": 10933,
      "task_loss": 0.05483207106590271
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01322152093052864,
      "epoch": 10.38,
      "learning_rate": 2.6357875294488865e-05,
      "loss": 0.0234,
      "step": 10934,
      "task_loss": 0.11500921845436096
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009567633271217346,
      "epoch": 10.38,
      "learning_rate": 2.6347235654478482e-05,
      "loss": 0.0093,
      "step": 10935,
      "task_loss": 0.006448997184634209
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014935207553207874,
      "epoch": 10.39,
      "learning_rate": 2.6336595769736245e-05,
      "loss": 0.0213,
      "step": 10936,
      "task_loss": 0.07824273407459259
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011560074985027313,
      "epoch": 10.39,
      "learning_rate": 2.6325955642194948e-05,
      "loss": 0.011,
      "step": 10937,
      "task_loss": 0.005575112998485565
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.031049327924847603,
      "epoch": 10.39,
      "learning_rate": 2.6315315273787428e-05,
      "loss": 0.0339,
      "step": 10938,
      "task_loss": 0.059728413820266724
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.06698919832706451,
      "epoch": 10.39,
      "learning_rate": 2.630467466644655e-05,
      "loss": 0.0754,
      "step": 10939,
      "task_loss": 0.15107224881649017
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014802630990743637,
      "epoch": 10.39,
      "learning_rate": 2.629403382210524e-05,
      "loss": 0.0185,
      "step": 10940,
      "task_loss": 0.051325224339962006
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.008546770550310612,
      "epoch": 10.39,
      "learning_rate": 2.628339274269645e-05,
      "loss": 0.0206,
      "step": 10941,
      "task_loss": 0.12929676473140717
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017933053895831108,
      "epoch": 10.39,
      "learning_rate": 2.6272751430153186e-05,
      "loss": 0.0208,
      "step": 10942,
      "task_loss": 0.04631277918815613
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03135647252202034,
      "epoch": 10.39,
      "learning_rate": 2.62621098864085e-05,
      "loss": 0.0291,
      "step": 10943,
      "task_loss": 0.008471904322504997
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009145300835371017,
      "epoch": 10.39,
      "learning_rate": 2.6251468113395465e-05,
      "loss": 0.0087,
      "step": 10944,
      "task_loss": 0.005033126100897789
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009342268109321594,
      "epoch": 10.39,
      "learning_rate": 2.6240826113047235e-05,
      "loss": 0.0093,
      "step": 10945,
      "task_loss": 0.008507819846272469
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0181414894759655,
      "epoch": 10.4,
      "learning_rate": 2.6230183887296955e-05,
      "loss": 0.0168,
      "step": 10946,
      "task_loss": 0.00474889762699604
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03423862159252167,
      "epoch": 10.4,
      "learning_rate": 2.6219541438077855e-05,
      "loss": 0.0467,
      "step": 10947,
      "task_loss": 0.15890000760555267
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.024610301479697227,
      "epoch": 10.4,
      "learning_rate": 2.620889876732317e-05,
      "loss": 0.0291,
      "step": 10948,
      "task_loss": 0.06941775232553482
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013289782218635082,
      "epoch": 10.4,
      "learning_rate": 2.6198255876966204e-05,
      "loss": 0.0125,
      "step": 10949,
      "task_loss": 0.005243049934506416
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026343289762735367,
      "epoch": 10.4,
      "learning_rate": 2.6187612768940293e-05,
      "loss": 0.0311,
      "step": 10950,
      "task_loss": 0.0735698863863945
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.015885502099990845,
      "epoch": 10.4,
      "learning_rate": 2.61769694451788e-05,
      "loss": 0.0194,
      "step": 10951,
      "task_loss": 0.051202740520238876
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.057844310998916626,
      "epoch": 10.4,
      "learning_rate": 2.616632590761514e-05,
      "loss": 0.0603,
      "step": 10952,
      "task_loss": 0.08287277817726135
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012808309867978096,
      "epoch": 10.4,
      "learning_rate": 2.615568215818276e-05,
      "loss": 0.0121,
      "step": 10953,
      "task_loss": 0.005606703460216522
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016954544931650162,
      "epoch": 10.4,
      "learning_rate": 2.6145038198815152e-05,
      "loss": 0.0158,
      "step": 10954,
      "task_loss": 0.005178598687052727
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0657939538359642,
      "epoch": 10.4,
      "learning_rate": 2.6134394031445843e-05,
      "loss": 0.0743,
      "step": 10955,
      "task_loss": 0.15127022564411163
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02318597212433815,
      "epoch": 10.4,
      "learning_rate": 2.6123749658008383e-05,
      "loss": 0.0273,
      "step": 10956,
      "task_loss": 0.06436937302350998
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.050585560500621796,
      "epoch": 10.41,
      "learning_rate": 2.6113105080436396e-05,
      "loss": 0.0512,
      "step": 10957,
      "task_loss": 0.05654662102460861
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.03527712821960449,
      "epoch": 10.41,
      "learning_rate": 2.6102460300663506e-05,
      "loss": 0.0369,
      "step": 10958,
      "task_loss": 0.0516989640891552
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011321203783154488,
      "epoch": 10.41,
      "learning_rate": 2.60918153206234e-05,
      "loss": 0.0106,
      "step": 10959,
      "task_loss": 0.00376252643764019
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014566227793693542,
      "epoch": 10.41,
      "learning_rate": 2.6081170142249773e-05,
      "loss": 0.0197,
      "step": 10960,
      "task_loss": 0.06615670025348663
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01640435680747032,
      "epoch": 10.41,
      "learning_rate": 2.607052476747639e-05,
      "loss": 0.0155,
      "step": 10961,
      "task_loss": 0.007749777287244797
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.0176126379519701,
      "epoch": 10.41,
      "learning_rate": 2.6059879198237026e-05,
      "loss": 0.0162,
      "step": 10962,
      "task_loss": 0.0036756154149770737
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.034288398921489716,
      "epoch": 10.41,
      "learning_rate": 2.6049233436465498e-05,
      "loss": 0.0426,
      "step": 10963,
      "task_loss": 0.1172315701842308
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.010387426242232323,
      "epoch": 10.41,
      "learning_rate": 2.6038587484095673e-05,
      "loss": 0.0171,
      "step": 10964,
      "task_loss": 0.07761921733617783
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016161680221557617,
      "epoch": 10.41,
      "learning_rate": 2.6027941343061412e-05,
      "loss": 0.024,
      "step": 10965,
      "task_loss": 0.0944659486413002
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.009701198898255825,
      "epoch": 10.41,
      "learning_rate": 2.6017295015296665e-05,
      "loss": 0.0091,
      "step": 10966,
      "task_loss": 0.0035239197313785553
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02417180687189102,
      "epoch": 10.42,
      "learning_rate": 2.600664850273538e-05,
      "loss": 0.0263,
      "step": 10967,
      "task_loss": 0.04507189989089966
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07505623996257782,
      "epoch": 10.42,
      "learning_rate": 2.599600180731155e-05,
      "loss": 0.0926,
      "step": 10968,
      "task_loss": 0.2502027750015259
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.059122804552316666,
      "epoch": 10.42,
      "learning_rate": 2.598535493095919e-05,
      "loss": 0.0613,
      "step": 10969,
      "task_loss": 0.08061361312866211
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01452677696943283,
      "epoch": 10.42,
      "learning_rate": 2.5974707875612357e-05,
      "loss": 0.021,
      "step": 10970,
      "task_loss": 0.07877064496278763
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.019884496927261353,
      "epoch": 10.42,
      "learning_rate": 2.5964060643205153e-05,
      "loss": 0.0246,
      "step": 10971,
      "task_loss": 0.06657460331916809
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013773739337921143,
      "epoch": 10.42,
      "learning_rate": 2.5953413235671688e-05,
      "loss": 0.0197,
      "step": 10972,
      "task_loss": 0.07330326735973358
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021291140466928482,
      "epoch": 10.42,
      "learning_rate": 2.594276565494611e-05,
      "loss": 0.0245,
      "step": 10973,
      "task_loss": 0.0534619465470314
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.038762666285037994,
      "epoch": 10.42,
      "learning_rate": 2.5932117902962616e-05,
      "loss": 0.0361,
      "step": 10974,
      "task_loss": 0.012404365465044975
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.007273940369486809,
      "epoch": 10.42,
      "learning_rate": 2.5921469981655415e-05,
      "loss": 0.007,
      "step": 10975,
      "task_loss": 0.004414750263094902
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.09941526502370834,
      "epoch": 10.42,
      "learning_rate": 2.591082189295876e-05,
      "loss": 0.0983,
      "step": 10976,
      "task_loss": 0.08790935575962067
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013018698431551456,
      "epoch": 10.42,
      "learning_rate": 2.590017363880691e-05,
      "loss": 0.0121,
      "step": 10977,
      "task_loss": 0.004249611869454384
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.014059076085686684,
      "epoch": 10.43,
      "learning_rate": 2.5889525221134192e-05,
      "loss": 0.0204,
      "step": 10978,
      "task_loss": 0.07782687991857529
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013228708878159523,
      "epoch": 10.43,
      "learning_rate": 2.5878876641874928e-05,
      "loss": 0.0125,
      "step": 10979,
      "task_loss": 0.005660973489284515
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.026601828634738922,
      "epoch": 10.43,
      "learning_rate": 2.5868227902963493e-05,
      "loss": 0.0369,
      "step": 10980,
      "task_loss": 0.12926051020622253
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05293282866477966,
      "epoch": 10.43,
      "learning_rate": 2.5857579006334282e-05,
      "loss": 0.0522,
      "step": 10981,
      "task_loss": 0.046048957854509354
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.07201385498046875,
      "epoch": 10.43,
      "learning_rate": 2.58469299539217e-05,
      "loss": 0.0862,
      "step": 10982,
      "task_loss": 0.21407830715179443
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.011810416355729103,
      "epoch": 10.43,
      "learning_rate": 2.5836280747660225e-05,
      "loss": 0.0173,
      "step": 10983,
      "task_loss": 0.0669318437576294
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.023423772305250168,
      "epoch": 10.43,
      "learning_rate": 2.5825631389484323e-05,
      "loss": 0.0259,
      "step": 10984,
      "task_loss": 0.04826189577579498
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.022933730855584145,
      "epoch": 10.43,
      "learning_rate": 2.58149818813285e-05,
      "loss": 0.033,
      "step": 10985,
      "task_loss": 0.12334097921848297
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.04635820910334587,
      "epoch": 10.43,
      "learning_rate": 2.5804332225127294e-05,
      "loss": 0.0514,
      "step": 10986,
      "task_loss": 0.09712212532758713
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.012980525381863117,
      "epoch": 10.43,
      "learning_rate": 2.579368242281527e-05,
      "loss": 0.02,
      "step": 10987,
      "task_loss": 0.08316182345151901
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017791984602808952,
      "epoch": 10.43,
      "learning_rate": 2.5783032476327007e-05,
      "loss": 0.0227,
      "step": 10988,
      "task_loss": 0.06660401821136475
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.017145544290542603,
      "epoch": 10.44,
      "learning_rate": 2.5772382387597128e-05,
      "loss": 0.0236,
      "step": 10989,
      "task_loss": 0.08195091038942337
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.018846768885850906,
      "epoch": 10.44,
      "learning_rate": 2.5761732158560263e-05,
      "loss": 0.0216,
      "step": 10990,
      "task_loss": 0.04650796204805374
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.021697044372558594,
      "epoch": 10.44,
      "learning_rate": 2.5751081791151083e-05,
      "loss": 0.0356,
      "step": 10991,
      "task_loss": 0.16030679643154144
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013004236854612827,
      "epoch": 10.44,
      "learning_rate": 2.574043128730428e-05,
      "loss": 0.0135,
      "step": 10992,
      "task_loss": 0.017738407477736473
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.02329951897263527,
      "epoch": 10.44,
      "learning_rate": 2.572978064895457e-05,
      "loss": 0.0298,
      "step": 10993,
      "task_loss": 0.08830928802490234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.020550910383462906,
      "epoch": 10.44,
      "learning_rate": 2.5719129878036686e-05,
      "loss": 0.0193,
      "step": 10994,
      "task_loss": 0.007566181942820549
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05504726618528366,
      "epoch": 10.44,
      "learning_rate": 2.5708478976485402e-05,
      "loss": 0.0543,
      "step": 10995,
      "task_loss": 0.04714515060186386
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05265543609857559,
      "epoch": 10.44,
      "learning_rate": 2.569782794623549e-05,
      "loss": 0.0528,
      "step": 10996,
      "task_loss": 0.05369632691144943
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.01432766579091549,
      "epoch": 10.44,
      "learning_rate": 2.5687176789221784e-05,
      "loss": 0.0136,
      "step": 10997,
      "task_loss": 0.0071462057530879974
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.016092218458652496,
      "epoch": 10.44,
      "learning_rate": 2.5676525507379097e-05,
      "loss": 0.0155,
      "step": 10998,
      "task_loss": 0.010227423161268234
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.013000791892409325,
      "epoch": 10.45,
      "learning_rate": 2.566587410264229e-05,
      "loss": 0.021,
      "step": 10999,
      "task_loss": 0.09294840693473816
    },
    {
      "compression/magnitude_sparsity/sparsity_level_for_model": 0.6249402080370299,
      "compression/magnitude_sparsity/sparsity_level_for_sparsified_layers": 0.8000000257231671,
      "compression/magnitude_sparsity/target_sparsity_level": 0.8,
      "compression_loss": 0.0,
      "distillation_loss": 0.05334077402949333,
      "epoch": 10.45,
      "learning_rate": 2.565522257694625e-05,
      "loss": 0.0488,
      "step": 11000,
      "task_loss": 0.0074908919632434845
    },
    {
      "epoch": 10.45,
      "eval_accuracy": 0.9128440366972477,
      "eval_loss": 0.42293068766593933,
      "eval_runtime": 17.8866,
      "eval_samples_per_second": 48.752,
      "eval_steps_per_second": 6.094,
      "step": 11000
    }
  ],
  "max_steps": 29484,
  "num_train_epochs": 28,
  "total_flos": 4.62792613049088e+16,
  "trial_name": null,
  "trial_params": null
}